Abstract
Existent literature helped narrow down variables influencing modern contraceptive adoption (usage), a behaviour carrying enormous positive externality. Using finite population sample size formula and probability proportional to size method of sample selection, primary data was collected from participants using inclusion and exclusion criterions. Binary logistic regression model was used to predict probability of occurrence of dependent variable ‘usage of modern method of contraception’ being treated at a dichotomous outcome level. Predictor variables after confirming association by cross tabulation were introduced stepwise to build model subject to elimination of those variables adding insignificantly to the overall predictability of the model. Variables such as gender, education level, spousal influence, extended family influence, financial well-being and contraceptive information were found to significantly predict the probability of occurrence of the dependent variable. Except for financial well-being with three sub-categories, other independent variables were treated at dichotomous level. Income level was found to be an important predictor although found statistically insignificant. Non-contributory factors such as age, occupation and years of marriage were dropped. Post-model construction, borrowing ‘nudges’ from behavioural economics (BE) domain, strategies to nurture the significant context specific influencing variables, were articulated. BE was particularly preferred for its openness to the paradigm of non-rational behavioural choices.
Keywords
Introduction
Population Reference Bureau (PRB) report 2019, makes India’s modern contraceptive prevalence rate (mCPR) at 47.7% while China, in comparison, stands at mCPR 84%. A fall-out from lack of contraception usage results in twin problems of high maternal and infant mortality rates which gets accentuated with ill-timed and unwanted pregnancies. Appropriate modern methods of contraception can effectively deal with issues arising out of financial stress, marital discords, infantile health complications, reproductive health related problems, sexually transmitted diseases, bulging family sizes, disoriented psychological state of mind, shrinking lifestyle standards and an incompetent suboptimal labour force.
Alano and Hanson (2018) verified using interpretative phenomenological qualitative methodology, on rural women in Southern Ethiopia, found out from narrated life experiences that the opportunities embarked upon once they adopted contraception, were enormous. We further analysed, Institute for Women’s Policy Research Report 2019 by Bernstein and Jones, who elaborately enumerated economic effects of access to contraception being improved educational attainment, labour force participation, career outcomes and lifestyles bettering future generations.
One hundred and sixty-nine member countries at International Conference on Population and Development; Program of action emphasised extending the access of modern methods of contraception so as to exercise the reproductive rights and ensure wellbeing of individuals belonging to reproductive age group under the 2030 Sustainable Development Agenda. In alignment majority of nations made amendments in budgets but in spite of a myriad of evidence of public investments and outreach programs by majority of nations there are also ample instances of bottlenecks in adoption of contraception due to various endogenous and exogenous factors influencing the behaviour. Common bottlenecks were psychological and economic factors leading us to look into the realms of behavioural economics (BE). Behavioural economics in reproductive health initiative (BERI) framework was exclusively studied for this purpose (Action, 2015).
Set in the Indian context, way back in 2006, curiosity of existing association between profile and contraception usage made Shah, Pradhan, Reddy and Joseph undertake a correlational study in the city of sub-urban Bangalore, state of Karnataka, India and reported age, education, years of marriage, occupation and religion linked to contraception usage. In 2011, a descriptive frequency based study by Hussain in the district of Malda, West Bengal, India, found that socio-economic factors such as income, education, wealth, place of residence, type of household, occupation, age of marriage, source of information about contraception, number of children, spousal influence and religion had a strong bearing on the contraception adoption. Not far in neighbouring Bangladesh, relationship of association led Haq et al. (2017) look into similar factors and find out the controlled influence of these factors by running a binary logistic regression on the same dependent variable. To have clarity on the methodology of analysis and interpretation we referred another study in Rwanda by Habyarimana and Ramroop (2018) using secondary DHS data. We found similar pattern of logistic regression analysis on the same dichotomous dependent variable. These articles gave us a strong foothold of the factors to be included and the statistical data handling procedure for our study.
Other than above quoted factors, multiple studies also reported that external variables such as peer pressure, societal expectations, culture, religious inclinations and advice of specific family members were crucial. We found clear evidence in Kulczycki (2008) reporting crucial importance of couple’s concordance in terms of communication and agreement while making contraceptive decisions. Using a multivariate logistic regression method, they arrived at the conclusion that not only couple cooperation and agreement is important for success of contraceptive usage but also that there is difference in reporting of contraception usage details by respective partners as individuals and as a couple. So, it can be fairly stated that data collected in surveys should focus on data gathered from individuals even though it pertains to married individuals and deals with spousal relationship equations. This point of data variability and methodology was kept as a reference for us in our data collection and analysis criteria.
Char et al. (2010) in their qualitative analysis of influence of mothers-in-law over young couples’ decisions on child bearing, in rural Indian households in the state of Madhya Pradesh, India, stated that this influence is more evident in the domains of ascertaining number of children, sex of children desired, contraception timing and so on but when it comes to reversible contraceptive choices the couples exercise more control in taking their own decisions. Thus, we started zeroing down on inclusion of specific family members maybe as an influence group for our study. This made natural sense especially in countries like India, where it is common to have joint extended families, who seem to wield dominant influence over contraceptive choices. To further strengthen this evidence, we looked into few more studies. Using multiple regression, Hajason et al. (2013), studied the influence of spousal and extended family dynamics among married couples in Madagascar, over dichotomous outcome of contraception adoption. They reported that contraceptive usage is directly related to spousal dynamics and even in some cases extended family agreement and amiability although the former is more strongly associated. Later, researchers also looked at their influence on well-being measured at four levels namely psychological, physical, intellectual and economic. This was interesting since we wanted to test financial wellbeing as a factor rather than an outcome of contraception usage. Using multinomial logistic regression another reviewed study set in Pakistan by Hameed et al. (2014) stressed on cohesive couple relationship, counselling and financial independence of partners for increasing contraception adoption rate. Adopting qualitative research methodology, study undertaken in rural Pakistan by Mustafa et al. (2015) emphasised contraceptive information, extended family influence and male involvement as important influencers.
The last component we wanted to examine was modern contraceptive choices. The literature we reviewed clearly favoured providing multiple contraceptive choices (Clark et al., 2018; Moreira et al., 2019; Muttreja & Singh, 2018; Sensoy et al., 2018). On the contrary, reviewing BE perspectives in the article by Ashton et al. (2015) under BERI we were curious to know further that whether the multiplicity of choices created confusion among potential contraceptive users. However, to our best knowledge, there was a gap of clarity in this kind of information as claimed by the above article as well. Therefore, we went ahead and included a section of questions in our questionnaire about whether they had received a single modern contraceptive choice or a multiple range of choices, as a follow-up of the health counselling they received. The other question that was asked was about whether or not it would be easier to adopt a modern method of contraception if presented as a single choice from a reliable source or rather as an array of choices to pick from, based on their preferences and needs.
Material and Method
Background and Area of Study
Karnataka (Mysore) a pioneer in 1931, was world’s first state to initiate clinics for contraception counselling (Rao, 1983). According to 2015–2016 National Family Health Survey (NFHS) Round 4, Karnataka has a CPR in the range of 60–64%. Positioned well among the four southern peninsular states of India according to a study on modern contraception usage in 2019 (Dey, 2019), it is interesting to find the juxtaposition of Karnataka. It lags behind in service deliveries, infrastructural quality, RH performance indicators, quality and monitoring, ease of financial transactions and OOPE among the southern states according to National Institution for Transforming India (NITI) Aayog (2015–2016 Report on Ranks of Larger States/South Indian States and Union Territories) under the aegis of World Bank and Ministry of Health and Family Welfare. We chose this state for this reason. In the second stage, district Ramanagara was chosen keeping in mind its unique features. Created out of another older district Bangalore Rural in 2007, this last district added to Karnataka displays a moderate rank 10 out of 30 districts. Classification based on relation between Economic Development and Human Development, 2011, puts it under Category 3 denoting ‘income and capability poor’. In terms of health indicators, it has mediocre infant, child and maternal mortality rates along with low per capita health expenditure. NFHS 4 2015–2016 puts the district mCPR at 56%–60% and District Level Household Survey (DLHS) 4 at 69%. This district made a meaningful choice for our study.
Sample Size
According to the last documented 2011 Ramanagara census, currently married females and males were enumerated at 72,449 and 20,110 respectively. This number was arrived from the census (See Appendix: Table C6: Age and Marital status structure of districts). The females who were 18 and above, along with men 21 and above with up to 20–29 years of marriage at the time of marriage were only included since it added up to 18–49 years of reproductive age group within marriageable age limit. Total number of married individuals stood at 92,609. Assuming decadal growth rate of population to be about 0.5% per annum (5.06% reported in 2001–2011), keeping a conservative estimate at about less than 1% annually, about 0.1 million married individuals. Out of these, those living in an ‘extended family’ (couple living with in-laws or relatives) model would be further fewer in number. Last but not the least, number of individuals who needed contraception (non-pregnant females and individuals who were not planning for their first and second child) would be even fewer.
The formula used for calculating sample size proportional to a finite population was:
N * X / (X + N – 1), where, X = Zα/22 * p * (1−p) / MOE2 and Zα/2 is the critical value of the Normal distribution at α/2 (e.g., for a confidence level of 95%, α is 0.05 and the critical value is 1.96), MOE is the margin of error at 5%, p is the sample proportion at 50%, and N is the population size at 100,000 individuals. The sample size estimated was 383 individuals. Assuming a response rate of 90%, 426 married individuals (still married and living together with their spouse and extended family at the time of survey under the Indian Law), living in ‘extended family’ type of household and needing contraception were chosen using multi-stage stratified sampling for the purpose of this survey. Four hundred and twenty-six individuals were included and barring one questionnaire, which was discarded because of incomplete response, all 425 questionnaires were included in the statistical analysis.
Sampling Method
Due to unavailability of village-wise data regarding married individuals under RH age group we used an approximate proxy by considering the number of households (HHs) as representative in a specific district. One participant was chosen from one HH. HHs were selected at cluster level based on Probability Proportional to Size (PPS) method, the details of which will be provided. Inclusion criterions were living in an extended family, married at a legal age, married for less than 30 years, currently married, in the RH age group of 19–49 (18 being the legal age for girls to get married whereas 21 is for boys and almost a year minimum to have a child after marriage), have one surviving child, did not have a pregnancy or childbirth in last 1 year and had used modern method of contraception in last 1 year for more than 6 months in continuation. In case both partners were present on the day of data collection it was left to them to decide who wanted to answer the questionnaires. They were left free to consult with each other. Selection of a married couple as two respondents was avoided from a single HH based on an earlier study showing reporting biases (Kulczycki, 2008).The specific HHs were chosen using a two-stage PPS method based on information provided by the District Census Handbook, Ramanagara, village and town wise Primary Census Abstract (PCA) Abstract, 2011. In the first stage all the villages and towns along with their total number of HHs were enumerated in the district of Ramanagara in accordance to their successive village and town codes. The total number of HHs in each respective village and town was cumulated and a cumulative aggregate figure of existing HHs was reached. There were 875 villages and towns with 204,456 HHs in total which became the primary sampling units (PSU). It was decided to have 25 clusters of 35 villages and towns in each to cater to the universe of 875 villages and towns (25 * 39 = 875). Sampling interval (SI) of 8,178 was arrived at by assigning 204,456 HHs over intended 25 clusters. A random start (RS) number 5,603 was chosen using a computerised array of random numbers between 1 and 8,178. Adding SI (5,603) to RS (8,178) the first cluster was selected with equal or greater than 13,781 HHs. The next cluster was arrived at by adding SI to 13,781 equating to 19,384, next was 24,987, followed by 30,590, 36,193, 41,796, 47,339, 53,002, 58,605, 64,208, 69,811, 75,414, 81,017, 86,620, 92,223, 97,826, 103,429, 109,032, 114,635, 120,238, 125,841, 131,444, 137,047 & 142,650 as the twenty fifth cluster. To reach a conservative estimate of 426 HHs, 17 units of HHs (25 clusters * 17 HHs = 425 HHs and one HH was chosen again from the last cluster) were chosen in each cluster.
As shown in Table 1, 17 HHs were chosen from each of the villages selected as clusters from Magadi, Ramanagara, Channapatna and 18 HHs from Kanakpura sub-district (taluka) in District Ramangara. The HHs in each cluster were chosen keeping the central post office as starting point and using google satellite location to identify peripheral residential areas. The central post office was universally found to be in a thickly populated residential area in all clusters. Data was collected during the period starting from May 2019 to December 2019.
Description of PPS Sampling
Ethical Clearance
Ethical clearance was obtained from the Research Ethics Committee of the affiliating University (CU: RCEC/01/03/19). Verbal consent was taken from the Head of the HH and each participant. Individuals were given prior orientation about the purpose and no risks involved in the study. Identifying information (like name, phone number, address etc.) was excluded from the questionnaire to ensure privacy and confidentiality. The right of individuals not to participate or drop off from the study was also respected.
Tool
The survey questionnaire was designed into four parts. First one catered to collection of demographic background information about age, gender, years of marriage, socio-economic variables such as occupation, education level and income level, second dealt with the family dynamics such as spousal influence, extended family influence on contraception decisions, third catered to number of available modern contraception choice information and last section was about financial wellbeing which was measured using Consumer Financial Protection Bureau (CFPB) financial wellbeing (FWB) standardised scale consisting of ten items gathering information regarding current financial status, financial independence and control over future financial status.
Variables
The dependent variable chosen for the study was the adoption of modern method of contraception (any method) which was treated as a dichotomous outcome (presence or absence of contraceptive use). The categorical predictors were dummy coded 0 and 1 for dichotomous classes except for FWB considered at three categories. Age was the only continuous predictor/independent variable included in the study. FWB scores were interval variables which were converted into ranges of low, medium and high and treated at ordinal level. Binary Logistic Regression was used to examine the influence of the predictors on the dependent variable. The categorical variables were first tested for meeting the assumptions of multicollinearity. Cross tabulation was used to check for associations before starting to construct the model. Predictors were introduced in steps to build the model and those variables which added insignificantly to the overall predictability of the model were removed and the model was rerun till the final model was reached. Three of the independent variables namely the ‘years of marriage’, ‘age’ and ‘occupation’ were dropped since they were not adding to the predictability of the model. The final model was derived with predictor variables of gender, education level, income level, spousal influence, extended family influence, contraception choice information and FWB.
Criterions for Logistic Regression
Since all predictors were categorical variables, so no need arises for checking linearity of the logit of predictor variable with the outcome variable. The multicollinearity (except for string Variable FWB) was tested for all 6 predictor variables. The results are reported in Tables 2 and 3. Table 2 indicated that the tolerance of none of the predictors was below 0.1 and no VIF being >10, confirming absence of multicollinearity. Further, Table 3 confirmed none of the predictors, as a pair, show a huge dependency percentage of variance proportion on a small eigenvalue so multicollinearity was ruled out.
Collinearity Statistics I
Collinearity Statistics II
Construction of the Model
The null model with only intercept was introduced first without any predictors. It indicated a prediction rate of 50.8% only (Table 4).
Null Model with Intercept Predictability
bThe cut value is 0.500.
Interpretation of the Null Model
Table 5 was used to derive the equation of regression of the ‘intercept only’ model. Ln (p/1−p) = β0 = 0.033; Probability of Contraceptive Users = exp (0.033)/1 + exp (0.033) = 1.0336/2.0336 = 0.508. SPSS calculated the probability of modern method of contraception usage for each individual using the block 0 model. If the probability of usage was 0.5 or more it would predict usage of contraception (as users =1) and no usage if the probability was less than 0.5. As more people used rather than not used, the probability of usage is 0.508 and therefore everyone is predicted as using modern method of contraception (coded as 1). As 50.8% of people were correctly classified, classification from the null model could be termed as 50.8% accurate. The addition of explanatory variables would increase the percentage of correct classification significantly if the model was good.
Intercept Significance without Predictors
Thereafter, each of the independent variable was introduced on a block basis, one variable at a time, using the ‘next’ and ‘enter’ functions of SPSS 20 version. The variables which were found non-contributory based on the omnibus test of model coefficient results at block level and model level, were removed from the model. The demographic variables of ‘years of marriage’, occupational’ and ‘age’ indicated no significant influence on the dependent variable and were not as such contributing to the model predictability. They were thereby dropped. Two-way interaction terms were also introduced to explore expected relationships influencing their way towards affecting the dependent variable to fall in one of the two groups of outcomes. None of them were found to contribute towards adding to the model predictability significantly. The final model was able to predict the overall outcome 79.8% of the times. The regression equation was derived by the final model coefficients.
Table 6 shows the description of all the categorical variables and their respective coding. Age, although excluded at a later stage, was the only continuous variable. The variable values were later converted into three age interval ranges for computing frequency. All the included variables except FWB was dichotomously treated.
Description of Variables in the Final Model
Table 7 shows the stepwise outcomes of the variables included in the final model. Each variable introduced added to the model predictability and the diminishing −2 Log likelihood showed the improvement. It can be seen that gender, education level, spousal and extended family influences, contraceptive information and level of FWB are significant predictors of modern methods of contraceptive adoption.
Stepwise Build-up of Final Model
Interpretation of the Variable Predictions in the Final Model
The ‘Variables in the Equation’ Table 8 summarises the importance of the explanatory variables individually while controlling for the other explanatory variables. p-Values of all the predictors are significant at confidence interval of 0.05 level when all other variables are controlled for except for the variable income level. That means there is no significant influence of income level on use of contraception. The column headed ‘B’ gives the coefficients of the predictors of the model. A negative value means that the odds of contraceptive users decreases for example: for those under low FWB and high FWB levels. Financial wellbeing (FWB) categories are treated as one compiled category first, then the low FWB is compared to medium FWB and finally high FWB with medium FWB which is the reference category. The table indicates that those under medium FWB are about twice as likely (1/0.516 = 1.937) to use contraception as compared to the low FWB individuals. Likewise, those belonging to medium FWB are 4.39 times (1/0.228 = 4.39) more likely to use contraception than individuals under high FWB. Females are 1.775 more likely to be users compared to males. School and beyond educated are 1.87 times more likely to belong to users. Individuals with ₹0.1 million –₹0.3 million annual income are 1.523 times more probable users. The individuals having spousal positive influence and agreement at personal level are 2.659 times more likely to use modern method of contraception. Those individuals with extended family perfect rapport are 3.16 times more predicted into users’ group. Those with single choice contraceptive information are 5.077 times more likely to be contraception users than those with multiple choice information.
Variables in the Final Model Equation
For our confidence interval, the fact that both limits are above 1 suggests that the direction of the relationship that we have observed is true in the population. If the lower limit had been below 1 then it would tell us that there is a chance that in the population the direction of the relationship is the opposite of what we have observed. The latter holds for the predictor confidence interval involving FWB, other things being controlled.
The classification of plots table at block 4 (entering until variable spousal influence) clearly indicates that the users are correctly predicted in majority times 70%–90% while non-users are predicted at lower rates in the range 50% and lesser. The column at the right side towards outcome 1 depicting users is much higher and populated in the classification plots. Then the trend in predicting correctly the outcome of users and non-users becomes equal with the introduction of extended family influence and contraceptive information. But, the final variable FWB although increasing the overall predictability of the model by a percent reduces its predictability to correctly determine users and weighs towards more correct in predicting non-users.
Interpreting Residuals
The main purpose of examining residuals is to (a) isolate points for which the model fits poorly, and (b) isolate points that exert an undue influence on the model. To assess the former we examine the residuals, especially the studentised residuals, standardised residuals and deviance statistics. To assess the latter, we use influence statistics such as Cook’s distance, DFBETA and leverage statistics.
(a) Studentised residuals and deviance statistics are all below 3 and only 6 cases in standardised residuals are between 3 and 4 which need inspection. This indicates model fits well except in these isolated case points. (b) DEFB0 1 and DFBETA rest of the values are all below 1 further proving no undue influence of predictors on the model. (c) Cook’s distance is all below 1 indicating there is no undue influence of predictors on the model. (d) Leverage should be between 0 and 1, in fact K + 1/N where k is number of predictors and N is sample size. Here k = 7 and N = 425, so K + 1/N = 0.0188. It’s in the range 0–1 and in the range 0.01–0.03, which shows no undue influence of predictors on the model.
Full Model being Tested
ln(p/1−p) = −3.937 + 0.574 x Female + 0.626 x School & Beyond Education Level + 0.421 x Annual up to 0.3 million Income Level + 0.978 x Complete Spousal Influence + 1.151 x Complete Extended Family Influence + 1.625 x Single Contraception Information − 0.661 x Low FWB −1.479 x High FWB
where,
xFemale = 1 for Gender, x School & Beyond Education Level =1 for Education Level, x Annual up to 0.3 million Income Level = 1 for Income Level, x Completely agreeable Spousal Influence = 1 for Spousal Influence, x Completely agreeable Extended Family Influence = 1 for Extended Family Influence, x Single Contraception Information = 1 for Contraception Choice Information, x Low FWB = 1 for FWB 1st class and x High FWB = 2 for FWB 2nd class
Discussion
Our findings indicate that the likelihood of contraceptive use was two-fold if the married individual belonged to medium range of FWB. The usage probability declined in lower and higher levels. Barring a few, the previous studies reviewed so far like by Hussain (2011), Dias and de Oliveira (2015), Haq et al. (2017) have generally indicated the usage as directly proportional to income and wealth endowments. However, our results were in alignment to the findings of Moreira et al. (2019) who cited reasons that the poorest lag because of lack of information and access whereas the richest due to less of health concerns and infrequency of sex. We suggest incentive techniques, working on loss aversion, to higher FWB group, stressing on health concerns which currently seem to be less probable although concerning to them. For the lower FWB group something like a ‘health buddy scheme’ wherein one oriented member for contraception adoption can be a reliable messenger to the rest of his/her acquaintances.
When it came to gender as a predictor of contraception use, Females were 1.78 times more probable than males supported by Armah-Ansah (2018). Do and Kurimoto (2012) had earlier suggested training women for better negotiating powers within partner relationships, which can be delivered only with improved financial independence. So, what could work as meaningful policies? One strategy could be behavioural nudge interventions like McConnell et al. (2018) suggested. The intervention varied between a standard voucher, a redeemable voucher with expiry deadline and last but not the least a voucher followed by a reminder SMS. The deadline and reminder nudge were found most effective. Kebede et al. (2019) concluded that demand for contraception is affected by information which was at best sub-optimal and male involvement needs to be targeted for effective use of contraception. But what was pulling back men from adopting contraception? Blackstone and Iwelunmor (2017) found that male attitude towards contraception was related to their perspective about female’s promiscuity and independence in making contraception decisions besides their self-profile. We inferred this as accentuating complications for men. We recommend a two-prong approach. First, men can be targeted for contraception adoption using the ‘affect’ approach of BE by powerful media usage like showing them as caring and responsible while taking both ‘cold and hot state decisions.’ Another way can be motivating women using ‘identity priming’ approach. Media and messages can portray women more often as sensible mothers who are interested in child spacing rather than naïve wives who fall prey to unprotected sex. These shifts in image can have a lasting impact in the minds of both male and female partners in a married relationship.
Our next predictor in the model was Education and the category of ‘school level and beyond’, was almost doubling the probability of contraception use. This result is in accordance to previous studies by Shah et al. (2006), Hussain (2011), Haq et al. (2017) and Abdulahi et al. (2020). The incentives approach can come in handy if participating in adult education can be rewarded with free or discounted contraceptive supplies. Educating will also help in making informed life choices. In fact, successful tools like lottery, a nudge grounded in BE, to promote contraception had been supported by Galárraga et al. (2018).
Next predictor in this study, income level, although wasn’t statistically significant at p-value of 0.05 yet was predicting 1.523 times to contraception usage at higher values of ₹0.1 million–₹0.3 million annual income range. Our findings were coherent with the findings of Hussain (2011). We recommend using ‘salience’ approach wherein the campaigns for adoption of contraception can actively project contraceptive products as contributing to changing living standards, for instance, bigger houses or female partners pitching in gainful employment opportunities which can actively boost their sales.
Another significant predictor was spousal influence, here defined as a marital condition where the married individuals talked, discussed and agreed on contraception usage decision, was leading to a 2.66 times increased predictability of adoption of contraception, an influence very crucial for adoption as emphasised by prior studies undertaken by Char et al. (2010), Hajason et al. (2013), Thulaseedharan (2018) and Abdulahi et al. (2020). In the context of especially low- and middle-income countries, another study undertaken by Phiri et al. (2015) found out by reviewing papers on BE that male involvement in initiation of contraceptive use and access to a free and respectable supply of contraception could hold key to better adoption and continuity rates of contraception by couples. We strongly recommend using ‘priming approach’ for pulling in both the partners in contraception awareness and advisory sessions. Male and female partners can be showcased as complements to each other and love could be symbolised by practice of contraception. Couple pictures on the face of the product packaging can lead to increase sales.
Our next finding was the influence of the extended family here defined as a family condition where the extended family members talked, discussed and agreed on contraception usage decision with the concerned married individuals. Contraception usage was being predicted with more than three times probability under supportive influence. This conformed to results found previously by Char et al. (2010) and Hajason et al. (2013). An intriguing fact was that many studies which emphasised extended family influence in fact looked into its negative contribution (Ghule et al., 2015; Gupta et al., 2014; Hussain, 2011). Contrarily, we looked into the constructive contribution. Making these members feel special may sort out a lot of misunderstandings and unnecessary blame games. Measures like bringing parents-in-law for regular awareness sessions followed with incentives like a certificate or a token of appreciation, can be extremely effective.
The final variable analysed was contraceptive information available, being single or multiple to the participants. According to paradigms of BE it’s way easier to inculcate a habit, if available choices are minimalistic. Testing on similar lines, we found that the frequency of contraception takers was much higher in those who had a single choice (provided by the health worker from nearest health centre, after assessing the individual’s need in all cases) rather than those provided with multiple choices. Single choice information increased model prediction by more than five times. Our claim is in total contrast to all our prior reviews. Each one reported expanded choice by providers and health workers would enhance contraceptive adoption. But, on the contrary, we found that if some reliable source recommended single optimal choice information tailored to the needs of concerned individuals, confusion was less and the adoption rate higher which was in alignment to the BERI framework article by Ashton et al. (2015). Reasons cited were actions dominated by habit, status-quo bias, complexity and limited attention. Solution suggested was a default contraceptive option, which is cognitively simpler to process and alters the existing lifestyle least. Stevens and Berlan (2014) pointed out that BE can come to rescue by offering tools of default with ‘opting out extension rather than opting in’ to increase the success rate of modern method of contraceptive adoption. Moreover, visibility of the most preferred outcome would ensure maximum hits. This solution suggested falls in line with the behaviourist paradigm of ‘limited attention’ which calls for heuristic solutions. We strongly suggest that after initial consultation with the medical personnel, one specific modern method of contraception should be prescribed. This might significantly increase adoption rate and reduce the complexities involved especially to individuals belonging to lower FWB, income and educational levels paving the way for desirable outcomes.
Conclusion
Karim et al. (2019) stated that interventions grounded in BE can go a long way in keeping up the motivation levels for continuing with contraception as planned because the drop-out rates of contraception was as high as the adoption rate marking the futility of public campaigns efforts. Most of the times individuals plan for their future, their finance, their wellbeing but implementation is a different ballgame altogether. So, where is the bottleneck? Well, the obstacle lies in those external variables which are often subtly contributing towards our decision making without us acknowledging or paying heed to them. For instance, elders of the family can either rightfully guide the potential family members about family planning with their insightfulness or mistakenly expose them towards ignorant beliefs. Wrong guidance can result in unplanned pregnancies and related complexities.
However, this is an unwarranted aspect in the domain of Economics, since it expects every individual to make informed rational choices. This is where the domain of Psychology brings in a rather new dimension in to Economics and offers a more realistic outlook towards human choices. BE, readily accepts diverse institutions surrounding our day-to-day lives, offering more humanistic approach-based solutions. Realistic solutions lie in modifying behaviour, without attempting to change lives by providing simple pointers, reminders, reinforcements or ‘nudges’ as they are called to serve as guides and motivators helping individuals to make more meaningful choices in alignment with their overall wellbeing (Dolan et al., 2012).
Our study is an effort towards this direction. It examines the influences surrounding contraception adoption and takes the path of BE to work out strategies to enhance adoption rate. The suggestions recommended try to set the stage in a manner that the rational mind takes a backseat and by default intended decisions happen heuristically.
Appendix.
Ever Married and Currently Married Population by Age at Marriage, Duration of Marriage
Footnotes
Acknowledgement
The study wouldn’t be possible without the contribution of the participants of the survey.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.
Funding
The authors received no financial support for the research, authorship and/or publication of this article.