Measuring Teacher Practices That Support Student Motivation: Examining the Factor Structure of the Teacher as Social Context Questionnaire Using Multilevel Factor Analyses

Abstract

We used multilevel factor analyses to investigate the structure of the Teacher as Social Context Questionnaire (TASCQ)-short form—a prominent measure of teacher practices that promote student motivation to learn. Based on Self-Determination Theory (SDT), the TASCQ contains three scales: Autonomy Support, Structure, and Involvement. Few studies have tested the construct validities of these three scales together. Furthermore, the few studies using factor analyses with these constructs’ scales showed mixed results. Moreover, none of those studies properly modeled the clustered nature of data with a multilevel analysis. We examined the structure of TASCQ scores from 697 fifth and sixth graders in 35 classes by conducting multilevel exploratory and multilevel confirmatory factor analysis, each with half the sample. Results indicated that students did not distinguish among Autonomy Support, Structure, and Involvement items; the best fit was a single factor.

Keywords

Self-Determination Theory teacher need-supportive practices multilevel exploratory factor analyses multilevel confirmatory factor analyses

Students’ motivation to learn in school is a crucial contributor to their academic success. Because teachers’ practices play an important role in fostering students’ motivation, researchers have paid considerable attention to teacher behaviors that are central in this process. According to Self-Determination Theory (SDT)—a dominant theory of motivation—teachers influence student motivation by supporting students’ basic needs of autonomy, competence, and relatedness (Deci & Ryan, 1985; Ryan & Deci, 2000). Teachers do this by encouraging student autonomy, providing structure for learning, and by being involved interpersonally (Connell & Wellborn, 1991; Reeve, 2002).

Researchers typically study teachers’ support of students’ needs with student surveys. In line with SDT, teaching practices are grouped into scales of Autonomy Support, Structure, and Involvement (e.g., Skinner & Belmont, 1993; Taylor & Ntoumanis, 2007). However, few studies have investigated the construct validities of these scales’ scores when all three types of need-supportive practices are considered simultaneously. Furthermore, results of those studies’ factor analyses differ. Moreover, none of these studies accounted for the clustered data (e.g., students in classrooms), which can bias the results (Huang & Cornell, 2016; Jöreskog & Sörbom, 2015). Therefore, in this study, we used multilevel factor analyses (MFA) to determine properly the factor structure of the clustered data of a common measure of teacher autonomy support, structure, and involvement—the Teacher as Social Context Questionnaire (TASCQ; Belmont, Skinner, Wellborn, & Connell, 1992).

Self- Determination Theory and Teachers’ Support of Students’ Needs

According to SDT, students’ motivation flourishes when their basic needs for autonomy, competence, and relatedness are satisfied (Ryan & Deci, 2000). Specifically, when students believe their behavior is self-determined or autonomous, they maximize their engagement, achievement, and well-being. In academic settings, three types of instructional practices (autonomy support, structure, and involvement) satisfy students’ basic needs, which promote student motivation (Reeve, 2002). Although each type of practice matches a psychological need (i.e., autonomy support with the need for autonomy, structure with the need for competence, and involvement with the need for relatedness), one type can help satisfy multiple needs (Taylor & Ntoumanis, 2007). For example, teacher autonomy support is associated positively with students’ perceived competence and sense of relatedness, in addition to autonomy (Jang, Reeve, Ryan, & Kim, 2009, Study 3).

Autonomy support helps students feel self-directed and autonomous. Autonomy-supportive practices include offering meaningful choices, emphasizing the relevance of content, providing rationales, listening to students, and encouraging initiative (Assor, Kaplan, & Roth, 2002; Reeve & Jang, 2006). Consistent with SDT, students’ perceptions of autonomy support are related positively to their autonomous motivation, interest, involvement, and achievement (e.g., Jang, Kim, & Reeve, 2012; Vansteenkiste et al., 2012).

Teachers’ provision of structure creates orderly, predictable, learning environments that help students focus on learning and develop their competence. Teachers do this by stating clear goals and directions, offering moderately challenging tasks, expressing high but attainable expectations, giving coherent explanations, and providing informative and consistent feedback (Jang, Reeve, & Deci, 2010; Stroet, Opdenakker, & Minnaert, 2015). Classroom structure is related positively to students’ autonomous motivation, use of learning strategies, and achievement (Hospel & Galand, 2016; Roth, Assor, Kanat-Maymon, & Kaplan, 2007; Skinner & Belmont, 1993).

Teachers’ involvement encourages students to feel connected to their school and classroom, and it promotes students’ positive relationships with teachers and other students (Anderman & Freeman, 2004; Niemiec & Ryan, 2009). Teachers communicate their involvement by fostering warm and respectful interactions, being attentive and fair to students, and investing effort in helping students learn (Haerens et al., 2013; Skinner & Belmont, 1993; Stroet et al., 2015). Teacher involvement is related positively to students’ motivation, feelings of belonging, interest, effort, attention, and achievement (King, 2015; Skinner & Belmont, 1993).

Considerable evidence indicates that teachers’ autonomy support, structure, and involvement foster many positive student outcomes. Hence, investigating teachers’ need-supportive practices is an important and active area of research.

The Structure of Teachers’ Need-Supportive Practices

Despite considerable researcher interest in teachers’ need-supportive practices, few studies have examined its structure empirically. In particular, the central premise that practices fall within one of three distinct types (i.e., autonomy support, structure, and involvement) has received little attention. Furthermore, the few studies using factor analyses show mixed results. Although some studies indicate that the three constructs are unique (e.g., Stornes, Bru, & Idsoe, 2008), other studies suggest they are not (e.g., Katz, Kaplan, & Gueta, 2009; Oga-Baldwin & Nakata, 2015).

Another issue regarding the structure of teacher need-support is that student data are nested in classrooms; therefore, factor analysis should account for both student and classroom levels. However, to our knowledge, only single level factor analyses—which can bias results of nested data—have been conducted (Jöreskog & Sörbom, 2015). Recent advances in measurement have shown how MFA can model clustered data more accurately, because it accounts for the greater within-class, compared with between-class, correlations among student responses (Jöreskog & Sörbom, 2015). Therefore, we assessed the distinctiveness of autonomy support, structure, and involvement using MFA.

The Teacher as Social Context Questionnaire (TASCQ)

The TASCQ (Belmont et al., 1992) is one of the most prominent measures of students’ perceptions of teachers’ autonomy support, structure, and involvement. It has been used in myriad contexts, including to measure learning environments of specific school subjects (e.g., Sierens, Vansteenkiste, Goossens, Soenens, & Dochy, 2009) and academics in general (e.g., Skinner & Belmont, 1993). Use of the TASCQ spans a wide range of grade levels, from elementary school (e.g., Skinner & Belmont, 1993) through college (e.g., Baeten, Dochy, & Struyven, 2013), and across continents (North America: Skinner & Belmont, 1993; Europe: Haerens et al., 2013; Asia: Taylor & Lonsdale, 2010). The TASCQ’s influence is not only due to its wide-spread use for measuring teacher need-supportive practices; however, it also guides the development of new instruments. Some TASCQ items are incorporated verbatim (e.g., Peetsma, Schuitema, & Van Der Veen, 2012), whereas others are adapted (e.g., Aelterman, Vansteenkiste, Van den Berghe, De Meyer, & Haerens, 2014), or serve as the basis for new items (e.g., Hospel & Galand, 2016). Thus, findings about the TASCQ’s factor structure have important implications for both conceptualizing and measuring practices that meet students’ needs and promote their intrinsic motivation.

The TASCQ comprises 52 items, grouped into four Autonomy Support subscales (Choice, Control, Respect, and Relevance), four Structure subscales (Contingency, Expectations, Help/Support, and Adjustment/Monitoring), and four Involvement subscales (Affection, Attunement, Dedication of Resources, and Dependability) (Belmont et al., 1992). Researchers often use the 24-item short version (Belmont et al., 1992); two items come from each of the long version’s subscales, making a total of eight items per construct. However, very few studies examined the factor structures of scores from all Autonomy Support, Structure, and Involvement items within either the short or long version of the TASCQ. Moreover, as we noted earlier, none used MFA.

We located only one study involving factor analysis of TASCQ item scores from the three subscales. The short form’s three-factor model was tested with confirmatory factor analysis (Lietaert, Roorda, Laevers, Verschueren, & De Fraine, 2015). Results showed a moderate but not excellent fit; however, the authors did not consider models with other numbers of factors.

In addition to the dearth of results about the TASCQ’s factor structure, results from other studies raise questions about the distinctiveness of its Autonomy Support, Structure, and Involvement subscales. Although the TASCQ is used often, as we noted previously, it is not typical for researchers to enter its three measures of need-supporting practices separately in analyses. In many cases, researchers created a composite or latent score of Teacher Need Support, in place of the separate subscales (e.g., Baeten et al., 2013; Skinner, Furrer, Marchand, & Kindermann, 2008; Van den Berghe, Tallir, Cardon, Aelterman, & Haerens, 2015). Furthermore, Vansteenkiste and his colleagues (2009, Study 2) reported results for both the three individual scales and the Need-Supportive composite.

None of the studies’ authors mentioned factor analysis when reporting their decision to combine scales into a composite. Correlations between the three subscales, however, showed that they are closely related to each other. Skinner and Belmont (1993) reported significant correlations ranging from .77 to .81. In other studies, researchers also noted high correlations between subscales: Autonomy Support and Structure (rs = .59-.71), Structure and Involvement (rs = .59-.67), Involvement and Autonomy Support (rs = .57-.72) (e.g., Lietaert et al., 2015; Taylor & Ntoumanis, 2007; Vansteenkiste et al., 2009). These high correlations among factors raise doubts about whether the subscales truly form three distinct factors; systematic examination of the factor structure of the TASCQ is needed.

The Current Study

Studies have not shown clearly that the TASCQ has a three-factor structure, with students perceiving teacher provision of autonomy support, structure, and involvement as distinct types of practices. Because the TASCQ is a widely used and influential instrument for measuring teachers’ need-supportive practices, examining its construct validity is crucial. Accordingly, we tested whether a three-factor structure (or some other structure) fitted the data well, using both multilevel exploratory factor analyses (MEFA) and multilevel confirmatory factor analyses (MCFA) (see Figure 1). In addition, we focused on the short form of the TASCQ because it comprises fewer items, and researchers use it more often than the long version.

Figure 1.

Proposed model of multilevel factor analysis of a three-factor model as TASCQ conceptualized.

Method

Participants

The participants were 697 elementary students (385 fifth graders, 312 sixth graders) from 35 classrooms in three elementary schools in Seoul, South Korea. There were 368 girls (53%) and 329 boys (47%). All students were Asian, aged 10 to 12 years old, and had active parental consent. Of the 35 teachers, six were males and 29 were females. The years of teaching experience varied substantially (M = 10.92 years, SD = 9.37 years).

Measures

The TASCQ (Belmont et al., 1992) measures students’ perceptions of their teacher’s need-supportive practices. The 24-item short form has three scales, each with eight items: (a) Teacher Autonomy Support (e.g., “My teacher gives me a lot of choices about how I do my schoolwork”), (b) Teacher Structure (e.g., “My teacher makes sure I understand before he or she goes on”), and (c) Teacher Involvement (e.g., “My teacher really cares about me”). Items are scored on a 5-point scale, ranging from 1 (not at all true) to 5 (very true). In our study, all items were specific to mathematics classes. We translated the TASCQ into Korean, and another person fluent in both Korean and English back-translated it into English for verification. The back-translated English version corresponded to the original TASCQ. Researchers using the original English version with upper-elementary students reported that the scale scores have acceptable internal consistency (αs = .76-.80; Belmont et al., 1992).

Procedure

Students completed surveys in their mathematics classes. We administered the surveys without the teacher present and assured students that their responses would be kept confidential. We collected the completed surveys at the end of the session.

Data Analysis

We first randomly split the sample in half using SPSS random sampling, to conduct separate exploratory and confirmatory factor analyses on distinct subsamples of student responses. Our sample had enough statistical power to warrant this division (Mundfrom, Shaw, & Ke, 2005; Tabachnick & Fidell, 2012). Each subsample showed no significant differences. We used MEFA and MCFA using Mplus (Muthén & Muthén, 2015) because the data were nested (students within teachers). There were too few schools (three) to model the school level (Chiu & Lehmann-Willenbrock, 2016). Because a sample of 50 at the highest level is often needed for stable parameter estimates in computer simulations, we do not make strong inferences about the 35 teachers (Maas & Hox, 2005).

MEFA

Because past studies have not provided clear evidence about the factor structure of the 24 original items of the TASCQ-short form, we conducted an MEFA with the first half of the sample. The MEFA identifies the variance at each level (Jöreskog & Sörbom, 2015). If there is substantial teacher-level variance, much of the differences in student responses occur between teachers, implying that students in the same class generally view their teacher similarly. In contrast, if nearly all the variance occurs at the student level, most of the differences are within teachers, implying that students in the same class have different views of their teacher. The factor loadings at the student level indicate how well an item accounts for the factor differences in student ratings of the same teacher. In contrast, factor loadings at the teacher level indicate how well an item accounts for factor differences in ratings across teachers. To allow flexibility in the possible factors, we used oblique rotations and report the one that accounts for the most variance.

To determine the number of factors in our MEFA, we examined the variance explained by each factor, and the ratios of adjacent factor’s eigenvalues (both formally and graphically with a scree plot). Evidence of a single factor includes the following: (a) explained variance exceeding 20% and (b) much higher ratio of the first factor’s eigenvalue over the second factor’s eigenvalue compared with other ratios of adjacent factors’ eigenvalues (Tabachnick & Fidell, 2012). Similarly, evidence for two factors entails two large eigenvalues that account for much of the variance, with small eigenvalues for the remaining factors (likewise three large eigenvalues for three factors, etc.).

Furthermore, we examined the factor loadings and factor correlation matrices (lambda and phi). We considered whether items within each construct have high loadings on the same factor (convergent validity) but not on different factors (no cross-loading); such results are evidence of a coherent factor (Tabachnick & Fidell, 2012). In contrast, items within a construct with low loadings on the same factor or substantial cross-loadings provide evidence that these items do not constitute a coherent factor. Also, low correlations (<.30) among factors in the phi matrix are evidence of distinct factors (Tabachnick & Fidell, 2012). In contrast, high correlations (>.70) among factors in the phi matrix raise concerns about the distinctiveness of each factor (Tabachnick & Fidell, 2012).

MCFA

Next, we conducted an MCFA with the second half of the sample to test the final model that was identified with the MEFA. We removed items whose factor loading was less than .45 in the MEFA (a fair cutoff threshold, compared with other possible thresholds: poor [.32], good [.55], very good [.63], excellent [.71]; Tabachnick & Fidell, 2012) or had cross-loadings of .45 or more. To assess the fit of the final model, we used the comparative fit index (CFI), Tucker–Lewis index (TLI), in addition to standardized root mean square residual (SRMR) and root mean square error approximation (RMSEA), which tend to minimize Type I and Type II errors under many conditions as shown in Hu and Bentler’s (1999) simulation studies. Fit thresholds are as follows: good (CFI & TLI > .95; SRMR < .08; RMSEA < .06), moderate (.90 < CFI & TLI < .95; .08 < SRMR < .10; .06 < RMSEA < .10), and poor (CFI & TLI < .90; SRMR > .10; RMSEA > .10).

Results

Before conducting MFA, we examined whether the items were normally distributed or not. One item (ST2) showed a nonnormal distribution across the sample sets (whole sample, MEFA subsample, MCFA subsample, respectively) with low skewness (–1.29, –1.34, –1.25) and high kurtosis (1.39, 1.83, 1.06), both more than |1|. Therefore, we did not include it in the following MFA (see Table A in Supplemental Material).

MEFA

The MEFA with a geomin oblique rotation of the first subsample (n = 347) showed that most of the variance is at the student level (90%) rather than the teacher level (10%); thus, the student level is primary, and students in the same classroom view the same teacher differently (see Table 1, footnote a). Other oblique rotations showed similar results. Correlations, variances, and covariances for each subsample are available from the first author on request.

Table 1.

Student- and Teacher-Level Variance, Eigenvalue, and Ratio (n = 347).

Factor	Student-level^a			Teacher-level^a
Factor	Variance^a	Eigenvalue	Ratio^b	Variance^a	Eigenvalue	Ratio^b
1	0.383	8.420	3.709	0.869	19.119	7.990
2	0.103	2.270	1.597	0.109	2.393	1.128
3	0.065	1.421	1.386	0.096	2.121	1.733
4		1.025			1.224

90% of the variance is at the student level, 10% is at the teacher level. The numbers in each % variance column refer to the proportion of variance explained at either the student or teacher level, respectively.

Ratio = Eigenvalue of this factor divided by that of the next largest factor.

The Factor 1 at the student level accounts for 38% of the variance, and its eigenvalue ratio over the Factor 2 is 3.7 (see Table 1 and Figure 2). In contrast, the corresponding values for the Factor 2 (10%; 1.6) and Factor 3 (7%; 1.4) are much lower. These results meet our criteria of a single dominant factor of more than 20% explained variance (38%) and a high ratio of adjacent eigenvalues (3.7 >> 1.6). The results at the teacher level strongly support a single, dominant factor (87% >> 20%; 8.0 >> 1.1). The corresponding teacher-level values for the Factor 2 (11%; 1.1) and Factor 3 (10%; 1.7) are much lower (for details, see Table B in Supplemental Material).

Figure 2.

Scree plot of multilevel exploratory factor analysis.

Next, we considered the factor loading and correlation matrices in each of the three-factor models (i.e., one-, two-, and three-factors; see Tables 2 -4). The two-factor and three-factor models’ variables do not fully load on the expected factors. In the two-factor model, item AS4 cross-loads on both factors. In the three-factor model, item IN1 cross-loads on Factors 1 and 3 at the student level, and item IN6 cross-loads on Factors 2 and 3 at the student level and Factors 1 and 3 at the teacher level. Furthermore, the phi matrices of the two-factor and three-factor models show substantial correlations among its student-level factors (two-factor: r = .59, p < .001; three-factor: rs = .58; .27; .47, ps < .01). Together, these results support a one-factor model.

Table 2.

Initial One-Factor Structure and Factor Loadings From Multilevel Exploratory Factor Analysis (n = 347).

Items	Factor loadings
In my class my teacher . . .	Student	Teacher
AS1 It seems like my teacher is always telling me what to do (R)	0.587	0.933
AS2 . . . doesn’t listen to my opinion (R)	0.675	0.921
AS3 . . . doesn’t give me much choice about how I do my schoolwork (R)	0.722	0.908
AS4 . . . is always getting on my case about schoolwork (R)	−0.005	0.699
AS5 . . . doesn’t explain why what I do in school is important to me (R)	0.460	0.990
AS6 . . . gives me a lot of choices about how I do my schoolwork	0.746	0.949
AS7 . . . listens to my ideas	0.750	0.977
AS8 . . . talks about how I can use the things we learn in school	0.673	0.847
IN1 . . . likes me	0.721	0.445
IN3 . . . really cares about me	0.759	0.914
IN4 . . . doesn’t understand me (R)	0.701	0.973
IN5 . . . spends time with me	0.480	0.986
IN6 . . . talks with me	0.647	0.934
IN7 I can’t depend on my teacher for important things (R)	0.579	0.925
IN8 I can’t count on my teacher when I need him or her (R)	0.527	0.986
ST1 Every time I do something wrong, my teacher acts differently (R)	−0.106	0.915
ST3 . . . keeps changing how he or she acts toward me (R)	0.401	0.924
ST4 . . . doesn’t make it clear what he or she expects of me in class (R)	0.616	0.909
ST5 . . . doesn’t tell me what he or she expects of me in class (R)	0.305	0.911
ST6 . . . makes sure I understand before he or she goes on	0.689	0.832
ST7 If I can’t solve a problem, my teacher shows me different ways to try to	0.700	0.948
ST8 . . . checks to see if I’m ready before he or she starts a new topic	0.741	0.976

Note. The model failed to converge when IN2 (my teacher knows me well) was included, so it was removed. Items in gray have factor loadings below .45. χ²(27) = 65.688, p < .001; CFI = .931; TLI = .972; SRMR = .042 (within) & .082 (between); RMSEA = .068. AS = Autonomy Support; IN = Involvement; ST = Structure.

Table 3.

Results of Multilevel Exploratory Factor Analysis of a Two-Factor Model (n = 347).

Item	Student-level		Teacher-level
Item	Factor 1	Factor 2	Factor 1	Factor 2
AS1	−0.299	−0.360	0.997	−0.163
AS2	0.668	0.120	0.901	0.099
AS3	0.704	0.119	0.998	0.048
AS4	0.531	0.495	0.672	0.158
AS5	0.362	0.167	0.943	−0.198
AS6	0.213	0.605	0.955	−0.019
AS7	0.119	0.693	0.918	−0.193
AS8	−0.022	0.726	0.867	−0.072
IN1	0.151	0.631	0.919	−0.014
IN3	0.015	0.796	0.932	−0.035
IN4	0.644	0.182	0.909	0.005
IN5	−0.176	0.680	0.871	0.212
IN6	−0.128	0.796	0.789	0.214
IN7	0.519	0.154	0.950	−0.006
IN8	0.561	0.061	0.982	−0.022
ST1	0.350	−0.414	0.000	0.901
ST3	0.688	−0.197	0.937	−0.024
ST4	0.710	−0.001	0.950	−0.266
ST5	0.422	−0.045	0.912	0.187
ST6	0.184	0.576	0.882	0.112
ST7	0.090	0.677	0.966	0.223
ST8	0.179	0.631	0.997	0.088

Note. The model failed to converge when IN2 was included, so it was removed. Items in gray have factor loadings below .45. Bold values indicate substantial cross-loading. χ²(27) = 86.970, p < .001; CFI = .897; TLI = .949; SRMR = .057 (within) & .143 (between); RMSEA = .080. AS = Autonomy Support; IN = Involvement; ST = Structure.

Table 4.

Results of Multilevel Exploratory Factor Analysis of a Three-Factor Model (n = 347).

Item	Student-level			Teacher-level
Item	Factor 1	Factor 2	Factor 3	Factor 1	Factor 2	Factor 3
AS1	0.272	0.389	0.009	0.999	0.137	0.006
AS2	0.726	0.007	0.104	0.903	0.081	−0.012
AS3	0.722	0.083	0.043	0.915	0.025	−0.141
AS4	0.428	−0.215	−0.313	0.630	0.173	0.281
AS5	0.199	0.527	−0.315	0.931	−0.174	0.010
AS6	0.161	0.650	0.030	0.982	−0.038	−0.220
AS7	0.241	0.369	0.374	0.998	−0.163	0.091
AS8	−0.127	0.821	0.044	0.823	−0.011	0.830
IN1	0.444	−0.011	0.589	0.905	−0.005	0.136
IN3	0.245	0.216	0.600	0.926	−0.032	0.021
IN4	0.649	0.170	0.020	0.912	−0.003	−0.037
IN5	−0.006	0.230	0.505	0.881	0.174	−0.021
IN6	0.002	0.438	0.416	0.745	0.225	0.568
IN7	0.496	0.199	−0.026	0.940	−0.003	0.071
IN8	0.539	0.117	−0.050	0.980	−0.024	−0.005
ST1	0.212	−0.075	−0.361	0.000	3.363	−0.002
ST3	0.753	−0.269	0.014	0.994	−0.068	−0.437
ST4	0.651	0.144	−0.143	0.926	−0.235	−0.018
ST5	0.408	0.005	−0.059	0.948	0.162	0.028
ST6	0.093	0.714	−0.066	0.911	0.095	0.020
ST7	0.085	0.628	0.107	0.978	0.149	−0.539
ST8	−0.009	0.974	−0.258	0.975	0.068	−0.073

Note. Items in gray have factor loadings below .45. Bold values indicate substantial cross-loading. χ²(25) = 98.507, p < .001; CFI = .772; TLI = .884; SRMR = .093 (within) & .134 (between); RMSEA = .093. AS = Autonomy Support; IN = Involvement; ST = Structure.

MCFA

Based on the results of the MEFA (one-factor model), we conducted an MCFA with a one-factor model using the second half of the sample (n = 350). One item (IN2) failed to converge and four items (AS4, ST1, ST3, ST5) had student-level factor loadings below .45 in the MEFA (see Table 2). As a result, we removed these five items before the MCFA. In the MCFA, the 18 items’ factor loadings at the student level all exceeded .45 (see Table 5). Furthermore, the fit indices showed a good fit (CFI = .968; TLI = .951; SRMR = .056 [within] & .061 [between]), and the RMSEA showed a moderate fit (.068). Likewise, the results showed high student-level reliability (.910) and teacher-level reliability (.919). Thus, a one-factor model shows a good-to-moderate fit to the data (see Figure 3). In addition, we tested a two- and three-factor model with MCFA, and the results support a one-factor model (see Tables C and D, Figures A and B in Supplemental Material).

Table 5.

Results of Multilevel Confirmatory Factor Analysis of a One-Factor Model (n = 350).

Item	Student-level			Teacher-level
Item	L	TD	SMC	L	TD	SMC
AS1	0.495	0.769	.242	0.437	0.402	.322
AS2	0.662	0.498	.468	0.286	0.038	.681
AS3	0.629	0.638	.383	0.283	0.233	.256
AS5	0.555	1.071	.223	0.292	0.731	.105
AS6	0.679	0.613	.429	0.428	1.301	.124
AS7	0.720	0.490	.515	0.246	0.031	.659
AS8	0.681	0.776	.374	0.325	0.100	.514
IN1	0.530	0.747	.273	0.113	0.056	.186
IN3	0.580	0.757	.308	0.500	0.891	.219
IN4	0.637	0.797	.338	0.748	1.948	.223
IN5	0.600	0.673	.348	0.505	0.379	.403
IN6	0.625	0.658	.373	0.687	0.768	.381
IN7	0.674	0.542	.456	0.294	0.100	.463
IN8	0.690	0.643	.425	0.400	1.212	.117
ST4	0.727	0.494	.517	0.234	0.060	.477
ST6	0.496	0.911	.213	0.136	0.739	.024
ST7	0.632	0.756	.346	0.181	0.072	.313
ST8	0.610	0.839	.307	0.147	0.074	.226

Note. The model failed to converge when IN2 was included, so it was removed. The factor loadings are unstandardized, and they can exceed 1. Items in gray have factor loadings below .45. χ²(304) = 557, p < .001; CFI = 0.968; TLI = 0.951; SRMR = .056 (within) & .061 (between); RMSEA = .068. L = Lambda or factor loading; TD = Theta-delta indicating measurement error; SMC = Squared multiple correlation; AS = Autonomy Support; IN = Involvement; ST = Structure.

Figure 3.

Final multilevel confirmatory factor analysis (n = 350) of a one-factor model of TASCQ with unstandardized factor loadings.

Four teacher involvement items had teacher-level factor loadings that exceeded .45 and hence account for factor differences in rating across teachers. These involved the teacher caring about, understanding, spending time with, and talking to students. These results suggest that students differentiate among teachers particularly with respect to items that reflect teacher involvement.

Discussion

We used MFA to examine the factor structure of the TASCQ-short form (Belmont et al., 1992)—a prominent measure of the three types of teacher instruction that supports students’ basic needs. Researchers have used the TASCQ widely for more than 25 years; the 24-item short form is particularly popular. Despite its prominence and its role in informing new measures (Aelterman et al., 2014; Hospel & Galand, 2016), few studies have formally examined the factor structure of its three scales. Furthermore, researchers have used TASCQ scale scores in different ways. Notably, however, none of the studies utilized MFA.

Our results indicate that students perceive the TASCQ-short form items as a single factor of teacher need-support, rather than differentiating among them to form unique factors communicating autonomy support, structure, or involvement. These results parallel others’ findings whereby middle graders did not differentiate among autonomy support, structure, and involvement practices (Katz et al., 2009; Oga-Baldwin & Nakata, 2015). Our results also provide validation for studies where researchers created a composite measure of teacher support (Skinner et al., 2008) or need support (Van den Berghe et al., 2015).

Our finding that the TASCQ’s items were perceived as a single dimension raises questions about the nature of other measures of teachers’ support for students’ needs. For example, how distinct, generally, are student-perceived autonomy support scores from teacher practices that provide structure or communicate involvement? Some of the items in the TASCQ address teacher behaviors that may arguably convey more than one type of need-supportive practices. For example, the Autonomy Support item “My teacher doesn’t listen to my opinion (reversed)” appears similar to Involvement (i.e., perceptions that the teacher is respectful, shows trust, and is attentive to students); it was also tied with other Involvement items on two- or three-factor structure (see Tables 3 and 4). There is little research that addresses the question of whether some types of practices represent more than one factor. Most SDT studies that focus on teacher practices measure autonomy support, but very few measure all three types of need support. Therefore, it is unknown whether some autonomy support items in other instruments also measure teacher structure and/or involvement. This information, added to our present results, may shed light on how methodologically and conceptually distinct teachers’ need-supportive practices are.

Most of the differences are across students rather than across teachers, indicating that students in the same classroom have different perceptions of the same teacher. This may reflect different teacher–student interactions around school work, perhaps because of students’ differing academic needs. In contrast, there was considerable agreement among students in the same class in terms of how they viewed their teacher’s involvement and efforts to build caring relationships.

Limitations and Future Research

Our use of only the short form of the TASCQ limits the generality of our results. Examining the full-length 52-item TASCQ is warranted, ideally using MFA. It may also be fruitful to consider the TASCQ’s factor structure with students from different grade levels, given that practices that are important for meeting the needs of early adolescents may not be as important for younger students or those in high school.

Conclusion

Although many studies use the TASCQ-short form to measure teacher support of students’ psychological needs, to date no study has examined its clustered data of students within classrooms with MEFA and MCFA. Our results showed that the students did not distinguish among TASCQ items designed to measure autonomy support, structure, and involvement. Hence, the best fit for the TASCQ scores was a single factor, namely one 18-item scale.

Supplemental Material

JPA_Supplementary_Materials_Submission_2_05-01-18_FINAL – Supplemental material for Measuring Teacher Practices That Support Student Motivation: Examining the Factor Structure of the Teacher as Social Context Questionnaire Using Multilevel Factor Analyses

Supplemental material, JPA_Supplementary_Materials_Submission_2_05-01-18_FINAL for Measuring Teacher Practices That Support Student Motivation: Examining the Factor Structure of the Teacher as Social Context Questionnaire Using Multilevel Factor Analyses by Inok Ahn, Helen Patrick, Ming Ming Chiu and Chantal Levesque-Bristol in Journal of Psychoeducational Assessment

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Supplemental Material

Supplemental material for this article is available online.

References

Aelterman

Vansteenkiste

Van den Berghe

De Meyer

Haerens

(2014). Fostering a need-supportive teaching style: Intervention effects on physical education teachers’ beliefs and teaching behaviors. Journal of Sport & Exercise Psychology, 36, 595-609.

Anderman

L. H.

Freeman

T. M.

(2004). Students’ sense of belonging in school. In Maehr

M. L.

Pintrich

P. R.

(Eds.), Advances in motivation and achievement (Vol. 13: Motivating students, improving schools: The legacy of Carol Midgley, pp. 27-63). Greenwich, CT: JAI Press.

Assor

Kaplan

Roth

(2002). Choice is good, but relevance is excellent: Autonomy-enhancing and suppressing teacher behaviours predicting students’ engagement in schoolwork. British Journal of Educational Psychology, 72, 261-278.

Baeten

Dochy

Struyven

(2013). The effects of different learning environments on students’ motivation for learning and their achievement. British Journal of Educational Psychology, 83, 484-501.

Belmont

Skinner

Wellborn

Connell

(1992). Teacher as Social Context: A measure of student perceptions of teacher provision of involvement, structure, and autonomy support (Technical report). Rochester, NY: University of Rochester. Retrieved from https://www.pdx.edu/psy/ellen-skinner-1

Chiu

M. M.

Lehmann-Willenbrock

(2016). Statistical discourse analysis: Modeling sequences of individual behaviors during group interactions across time. Group Dynamics: Theory, Research, and Practice, 20, 242-258.

Connell

J. P.

Wellborn

J. G.

(1991). Competence, autonomy and relatedness: A motivational analysis of self-system processes. In Gunnar

M. R.

Sroufe

L. A.

(Eds.), The Minnesota symposium on child psychology (Vol. 22, pp. 43-77). Hillsdale, NJ: Lawrence Erlbaum.

Deci

E. L.

Ryan

R. M.

(1985). Intrinsic motivation and self-determination in human behavior. New York, NY: Plenum Press.

Haerens

Aelterman

Van den Berghe

De Meyer

Soenens

Vansteenkiste

(2013). Observing physical education teachers’ need-supportive interactions in classroom settings. Journal of Sport & Exercise Psychology, 35, 3-17.

10.

Hospel

Galand

(2016). Are both classroom autonomy support and structure equally important for students’ engagement? A multilevel analysis. Learning and Instruction, 41, 1-10.

11.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1-55.

12.

Huang

F. L.

Cornell

D. G.

(2016). Using multilevel factor analysis with clustered data: Investigating the factor structure of the Positive Values Scale. Journal of Psychoeducational Assessment, 34, 3-14.

13.

Jang

Kim

E. J.

Reeve

(2012). Longitudinal test of self-determination theory’s motivation mediation model in a naturally occurring classroom context. Journal of Educational Psychology, 104, 1175-1188.

14.

Jang

Reeve

Deci

E. L.

(2010). Engaging students in learning activities: It is not autonomy support or structure but autonomy support and structure. Journal of Educational Psychology, 102, 588-600.

15.

Jang

Reeve

Ryan

R. M.

Kim

(2009). Can self-determination theory explain what underlies the productive, satisfying learning experiences of collectivistically oriented Korean students? Journal of Educational Psychology, 101, 644-661.

16.

Jöreskog

K. G.

Sörbom

(2015). LISREL 9.2. New York, NY: Scientific Software International.

17.

Katz

Kaplan

Gueta

(2009). Students’ needs, teachers’ support, and motivation for doing homework: A cross-sectional study. Journal of Experimental Education, 78, 246-267.

18.

King

R. B.

(2015). Sense of relatedness boosts engagement, achievement, and well-being: A latent growth model study. Contemporary Educational Psychology, 42, 26-38.

19.

Lietaert

Roorda

Laevers

Verschueren

De Fraine

(2015). The gender gap in student engagement: The role of teachers’ autonomy support, structure, and involvement. British Journal of Educational Psychology, 85, 498-518.

20.

Maas

Hox

(2005). Sufficient sample sizes for multilevel modeling. Methodology: European Journal of Research Methods for the Behavioral and Social Sciences, 1, 86-92.

21.

Mundfrom

D. J.

Shaw

D. G.

T. L.

(2005). Minimum sample size recommendations for conducting factor analyses. International Journal of Testing, 5, 159-168.

22.

Muthén

L. K.

Muthén

(2015). Mplus: The comprehensive modeling program for applied researchers: User’s guide (7th ed.). Los Angeles, CA: Author.

23.

Niemiec

C. P.

Ryan

R. M.

(2009). Autonomy, competence, and relatedness in the classroom: Applying self-determination theory to educational practice. Theory and Research in Education, 7, 133-144.

24.

Oga-Baldwin

Nakata

(2015). Structure also supports autonomy: Measuring and defining autonomy-supportive teaching in Japanese elementary foreign language classes. Japanese Psychological Research, 57, 167-179.

25.

Peetsma

Schuitema

Van Der Veen

(2012). A longitudinal study on time perspectives: Relations with academic delay of gratification and learning environment. Japanese Psychological Research, 54, 241-252.

26.

Reeve

(2002). Self-determination theory applied to educational settings. In Deci

E. L.

Ryan

R. M.

(Eds.), Handbook of self-determination research (pp. 183-203). New York, NY: The University of Rochester Press.

27.

Reeve

Jang

(2006). What teachers say and do to support students’ autonomy during a learning activity. Journal of Educational Psychology, 98, 209-218.

28.

Roth

Assor

Kanat-Maymon

Kaplan

(2007). Autonomous motivation for teaching: How self-determined teaching may lead to self-determined learning. Journal of Educational Psychology, 99, 761-774.

29.

Ryan

R. M.

Deci

E. L.

(2000). Self-determination theory and the facilitation of intrinsic motivation, social development, and well-being. American Psychologist, 55, 68-78.

30.

Sierens

Vansteenkiste

Goossens

Soenens

Dochy

(2009). The synergistic relationship of perceived autonomy support and structure in the prediction of self-regulated learning. British Journal of Educational Psychology, 79, 57-68.

31.

Skinner

E. A.

Belmont

M. J.

(1993). Motivation in the classroom: Reciprocal effects of teacher behavior and student engagement across the school year. Journal of Educational Psychology, 85, 571-581.

32.

Skinner

E. A.

Furrer

Marchand

Kindermann

(2008). Engagement and disaffection in the classroom: Part of a larger motivational dynamic? Journal of Educational Psychology, 100, 765-781.

33.

Stornes

Bru

Idsoe

(2008). Classroom social structure and motivational climates: On the influence of teachers’ involvement, teachers’ autonomy support and regulation in relation to motivational climates in school classrooms. Scandinavian Journal of Educational Research, 52, 315-329.

34.

Stroet

Opdenakker

Minnaert

(2015). What motivates early adolescents for school? A longitudinal analysis of associations between observed teaching and motivation. Contemporary Educational Psychology, 42, 129-140.

35.

Tabachnick

B. G.

Fidell

L. S.

(2012). Using multivariate statistics. San Francisco, CA: Pearson.

36.

Taylor

I. M.

Lonsdale

(2010). Cultural differences in the relationships among autonomy support, psychological need satisfaction, subjective vitality, and effort in British and Chinese physical education. Journal of Sport & Exercise Psychology, 32, 655-673.

37.

Taylor

I. M.

Ntoumanis

(2007). Teacher motivational strategies and student self-determination in physical education. Journal of Educational Psychology, 99, 747-760.

38.

Van den Berghe

Tallir

I. B.

Cardon

Aelterman

Haerens

. (2015). Student (dis)engagement and need-supportive teaching behavior: A multi-informant and multilevel approach. Journal of Sport & Exercise Psychology, 37, 353-366.

39.

Vansteenkiste

Sierens

Goossens

Soenens

Dochy

Mouratidis

. . . Beyers

(2012). Identifying configurations of perceived teacher autonomy support and structure: Associations with self-regulated learning, motivation and problem behavior. Learning and Instruction, 22, 431-439.

40.

Vansteenkiste

Sierens

Soenens

Luyckx

Lens

(2009). Motivational profiles from a self-determination perspective: The quality of motivation matters. Journal of Educational Psychology, 101, 671-688.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.69 MB