In our week 8 assignment the objective is to Conduct an appropriate analysis to determine, what, if anything, predicts self-efficacy in Statistics.
According to Bandura (1997) there is a correlation between our beliefs and completing a specific task, suggesting self-efficacy plays a major role and is highly influential on a person’s ability to achieve a goal, Bandura discussed a relationship between variable factors and four key principles which play a major role in accomplishment. Though is it the chicken or the egg, is it a question of competence or confidence. If I am competent in a task I am more likely to be confident. And believe I can do it. If I am not competent I am more likely to be less confident. Bandura ( 1997) did suggest even before we take any action there is a correlation of key principles. Though other factors to take into consideration are past experience and the link between past experiences and present learning. So for example if I am a competent driver, and I am travelling to a new destination I may be a be a bit apprehensive about my journey, though once I have travelled the journey on a few occasions it becomes much easier and I become more confident. Equally if a learner was going to start driving lessons, though had never been in a car before, it would be a far greater challenge one would think than if he or she had travelled on a car. One could also suggest past schemas play a role in self-efficacy, say for example you had built a negative association to delivering presentations in your youth, this may manifest late in life. So one may have the competence, though not the confidence. Though there is a fine line between confidence and competence, and our ability to learn something new is a correlation to many variable factors, including how the information is presented and our learning style.
Looking at previous research of predictors in self-efficacy and academia, by exploring the following research by Hall and Ponton,2002. We can see how in this comparative design study of two groups of students, briefly the study explores self-efficacy in maths, utilising a psychometric test designed to measure self-efficacy in maths, the study explores different levels in maths. Notably Algebra and calculus. Using an independent T Test the research found that the higher level maths group calculus, had a high degree of self-efficacy than the lower level maths group. So if we take Banduras ( 1997) theory into consideration it would suggest self-efficacy is linked to achievement. Though whilst the research did find some interesting correlations. Some of the things to take into consideration is the motivation to succeed, and the bigger picture, was the subject a component of additional study or course. Say for example in the mental health psychology course we cover statistics, having a certain level of knowledge of statistics is important, though not everyone who is doing the course will want to be a statistics teacher, given as such would require a different level of knowledge than if someone was choosing a different career path in the field, equally there are different components of a subject. And further more reflecting back on the research the aspirations of the group studying the higher level maths, would play a part in self-efficacy and past experienced, if one had a goal of being an engineer than knowledge of maths would be more significant than if one had aspirations of being a historian.
One of the features of the following research by Tiyuri et al. (2018) was it included multiple linear regressions to study the impact of self-efficacy and performance in academia, over 300 learners participated in the study, across a broad range of academic levels, from degree to PHD. The group was divided into pretty much an equal split between female and male. The gender (P = 0.754) and school P value (P = 0.364) showed non-significant levels, although the psychometric test did not demonstrate a significant difference based on gender and school, there was a significance difference (r = 0.393, P = 0.0001 in the level of academia amongst PHD students. Equally in this study one could imply that the students doing the PHD had more motivation than the other students, also it is important to note the correlation between past experiences, prior knowledge. So whilst this research did produce interesting findings of the link between self-efficacy and learning performance, the objective of the study was not to examine variables which correlated to act as a catalyst for what underlies the self-efficacy.
According to Field (2012).in order for testing to be accurate, we must make the assumption the sampling distribution is normal” the implications as such and methodology if there any doubts.
The reason for the analysis in the assignment:
The Rationale
In order to predict self-efficacy in statistics I conducted a standard multiple regression to predict self-efficacy in statistics we explored self-efficacy, confidence, usefulness, gender, attitudes, age, maths and science levels.
Regression analysis was used to determine which amongst the following variables self-efficacy, confidence; usefulness, gender, attitudes, and age correlate with maths and science to form self-efficacy in statistics. The regression helps us to analyse and understand which is the variables is statistically significance and non-significant and influential in the dependent variable. The spss output helps us to determine correlative factors. And isolate the variables and negative and positive relationships between variables, which factors are down to chance and which factors appear to be significant. For example the opportunity to examine p values and be able to test the hypothesis.
The regression analysis helped to isolate any patterns , and variable factors which colluded to determine this case the self-efficacy in statistics.
The reason I chose a Standard multiple regression was because i wanted to predict the values on the research date data establishing the continuous dependent variable factors such as ,Confidence, Usefulness ,Male Dominated Field, Tutor Attitudes and self-efficacy the independent variable, as well as the categorical variables such as age and maths and sciences.
The main aims of using the multiple regression in the analysis were to examine and analyse the effect of ,Confidence, Usefulness, Male Dominated Field, Tutor Attitudes on self-efficacy in statistics. Also the correlation of self-efficacy on Confidence,Usefulness,Male Dominated Field, Tutor Attitudes , how does a change in factor impact another.
This in turn gives us an opportunity to gain point estimates, understand and analyse correlative factors and predict further hypothesis.
The task was to identify self-efficacy in Statistics, I ran a multiple regression comparison on the Stats Confidence file provided
I decided to run a Standard multiple regression to predict predicts self-efficacy in Statistics. With the guidance of (SPSSVideoTutor.com 2014) and Dolan, 2018.
The date consisted of the following variables, Age, Self Efficacy,Confidence,Usefulness,Male Dominated Field, Tutor Attitudes and grades in grades in Math or Science.
Also important to note according to lens 2014 a regression analysis cannot be performed on ordinal level data, because maths and science are in this instance ordinal level data, I had to create dummy variables, which is something I really struggled with.
The study is, looks at the relationship between the variables as per above Self Efficacy,Confidence,Usefulness,Male Dominated Field,Tutor Attitudes grades in maths and science by participant’s to predicting any variables or correlations of variables of self-efficacy in Statistics.
The results of the output the are as follows
Standard multiple regression
R=.568a the value of R Square= .279 it appears the variables are not good predictors of self-efficacy in statistics
The ANOVA significance value of .000b suggests that one of the predictor variables, with this information as per the video Dolan (2018) we can confirm that minimum of one variable predicts self-efficacy in statistics
A multiple regression was run to predict the following variables self-efficacy from
- Confidence
- Usefulness
- Male dominated field
- Tutor attitudes –
- Maths
- Science
Following the ANOVA output These variables statistically significant predicated self-efficacy, F (6, 93)= 7.387, P <.000, R2=.323
The Standardized Coefficients for the following variables are
| Confidence -.339 |
| Usefulness -.146 |
| Male Dominated Field -.058 |
| Tutor Attitudes -.132 |
| maths -.028 |
| Science -.013 |
The significance values are
| Confidence .024 |
| Usefulness .221 |
| Male dominated field .529 |
| Tutor attitudes .294 |
| Maths .755 |
| Science .884 |
It appears that only one variable in the Confidence in the coefficients with a significance value of .024, significantly predicts self-efficacy, the rest of the variables in the coefficient are non-significant
The VIF values are
| Confidence 3.004 |
|
Usefulness 1.932 |
| Male dominated field 1.154 |
| Tutor attitudes 2.130 |
| Maths 1.117 |
| science 1.053 |
All of the VIF values are under 4 or 5, according to Dolan ( 2018) if the The VIF values are above 4 or 5, there could potentially be a problem with multi linearity
The Beta values
| Confidence -.339 | |
| Usefulness -.146 | |
| Male dominated field -.058 | |
| Tutor attitudes -.132 | |
| Maths .028 | |
| Science -.013 |
According to Dolan ( 2018) the higher the beta value, “the higher the impact of the independent variable on the dependent variable”. Interestingly enough, confidence has a beta value of -.339 the highest beta value and a p value of Confidence .024 the lowest significance value
Anova table
| df | F | R2 | P | ||
| Self-Efficacy | 7.387 | 6 | R2=.323 | P <.000 | |
Self-Efficacy
Following the ANOVA output These variables statistically significant predicated self-efficacy, F (6, 93)= 7.387, P <.000, R2=.323
A multiple regression table was run to predict self-efficacy from age , which produced the following significance values
| B | SE B | β | t | p | Sig. | ||||
| Confidence | -.339 | .024 | |||||||
| Usefulness | .146 | .221 | |||||||
| Male dominated field | -.058 | .529 | |||||||
| Tutor attitudes – | -.132 | . .294 | |||||||
| Maths | Nominal (dummy ) | .028 | .755 | ||||||
| Science | Nominal
( dummy) |
-.013 | .884 | ||||||
| Age | categorical | ||||||||
| Self-Efficacy | Independent variable | ||||||||
Conclusion:
Self-efficacy is abstract concept, confidence had a beta value of -.339 the highest beta value and a p value of Confidence .024 the lowest significance value, though how do you define confidence. And then there is the issue of confidence and competence, Quantitative states facts for example maths and science, Qualitative is more speculative in this instance. Reflecting on my own experience, journey, the module as a whole and this assignment. One of the areas I got bogged down on was the dummy statistics, during which I had a moment of introspection and reflection, I was beginning to think I had come so far in the module and gradually through the attainment of grades and tutorials and assignments I had built a conviction I could get through the module, though more importantly develop my skills around statistics, the challenge I had was not in the understanding though in the technical knowledge around creating the dummy variables, ad all I could think of was the weight of the final project and how a process that is relatively simple to do, could undo all the work I had done. Reflecting on the research papers I had read around the university students and the college students around maths, I was beginning to release self efficacy is a complex area to define. And whilst agree with some of the research by Bandura (1997), this assignment suggested the variables work as a cocktail or perfect storm, that collude. Qualitative Research can be challenging to define, and quantify, Quantitative data gives us values. Though there is a saying in sport, you do not play on paper, certain data can point you in a certain direction, though what is concise for one person may not be concise for someone else. And how you quantify Qualitative research can be a challenge in in itself. I think to a point all of the following variables we examine have an impact on self-efficacy on some level, although the data points towards confidence, this could be seen as a grey area. As you can start an assignment with to much confidence and that in itself could have negative implication, or you could start with little confidence, though that could grow as you go along. Though not journey is linear, each journey is filled with twists, turns, and permeations.
References:
Bandura, A. (1997). Self-Efficacy: The exercise of control. New York, NY: W. H. Freeman.
Field, A. (2012). The linear model (regression part I) [Video file]. Retrieved from https://www.youtube.com/watch?v=LZtPDgoskfI
Hall, M., & Ponton, M. (2002). A comparative Analysis of Mathematics Self-Efficacy of Developmental and Non-developmental Freshman Mathematics Students. Presented at the 2002 Meeting of Louisiana/Mississippi Section of the Mathematics Association of America.
Laureate Education (Producer). (2014). Weekly lecture notes, Week 8 – Part 1 [Video file]. Baltimore, MD: Author.
Laureate Education (Producer). (2014). Weekly lecture notes, Week 8 – Part 2 [Video file]. Baltimore, MD: Author.
SPSSVideoTutor.com (Producer). (n.d.). Multiple regression [Video file]. Deerfield Beach, FL: ConsumerRaters. Retrieved 28 October 2014.
Tiyuri, A., Saberi, B., Miri, M., Shahrestanaki, E., Bayat, B. B., & Salehiniya, H. (2018). Research self-efficacy and its relationship with academic performance in postgraduate students of Tehran University of Medical Sciences in 2016. Journal of Education and Health Promotion, 7, 11. http://doi.org/10.4103/jehp.jehp_43_17