Middle school students to college students are certainly familiar with research. Research is presented to train students and students to think scientifically. In writing scientific papers, we must know the hypothesis of the research. So, this hypothesis is the researcher’s preconceived notions of the problem to be studied. But understanding the hypothesis is not this simple.
The hypothesis comes from the Greek words hupo and thesis . Hupo is provisional, while thesis is a statement or theory. It can be concluded that the meaning of the hypothesis is a temporary statement. This is the researcher’s presumption of the research problem. However, this hypothesis is not the truth. Because of preconceived notions, hypotheses can be right or wrong.
The use of hypotheses, for example, is the process of research on the relationship between human habits of throwing garbage and the high amount of waste in Indonesia. Based on the temporary data that you get, the hypothesis that emerges is that human behavior is related to the amount of waste. That is, bad human habits affect the high amount of waste from time to time.
Hypothesis writing cannot be called the truth. Even if you design a hypothesis based on valid and strong data. To prove this hypothesis is true or not, you have to do the research. The results of the research will show whether it is in accordance with the hypothesis or actually produces new findings.
In several opinions, one of them from Zikmunda reveals that a hypothesis is a proposition or conjecture that has not been proven. So, the hypothesis is still tentative. Hypothesis statements only explain phenomena and possible answers to research questions. The real answer is obtained after the research is done.
Suryabrata, a writing expert, provides an explanation of the hypothesis in several ways. technically, a hypothesis is defined as a statement regarding the state of the population to be tested or studied. This study is based on data taken from research samples. Meanwhile, a statistical hypothesis is a statement regarding the state of the parameters tested through statistical samples.
Definition of Research Hypothesis
General Definition of Research Hypothesis
The hypothesis is generally interpreted as a temporary answer (conjecture) of a research problem. The hypothesis is only arranged in the type of inferential research, namely the type of research with a quantitative approach that aims to test. Testing a hypothesis is always through inferential statistical analysis techniques, while descriptive research does not require an explicit hypothesis formulation.
Hypotheses can be prepared by researchers based on a strong theoretical basis and supported by relevant research results. Researchers must understand about the content and how the steps in formulating a research hypothesis.
The formulation of the hypothesis has requirements or characteristics that must be met by the researcher. As for some of the characteristics of the formulation of the hypothesis, according to Soesilo (2015) as follows:
- The hypothesis is stated in a declarative statement , not a question sentence. This statement is the view of researchers based on the results of the study of the theory used.
- Researchers must be consistent (not changeable) about the contents of the hypothesis. Therefore, researchers need to conduct an in-depth study of the theory used in compiling the hypothesis.
- In experimental research, the hypothesis contains statements regarding the effectiveness, difference, or influence of one variable on another. In the hypothesis there are at least two variables studied.
- The hypothesis must be testable ( testable ). In addition to explaining the method (technique) of measuring each variable to be studied, the research methodology section also explains the analytical techniques used to test the research hypothesis.
Definition of Research Hypothesis According to Experts
1. The American Heritage Dictionary
The American Heritage Dictionary defines a research hypothesis as a provisional explanation of a scientific phenomenon that needs to be tested by further research. In other words, from this understanding we can conclude that a scientific hypothesis must be scientifically proven, and vice versa.
Kerlinger wrote that a hypothesis is a statement or conjecture based on two or more variables.
Suryabrata argues that the hypothesis used in quantitative research uses the deduction method. On the other hand, in qualitative research a hypothesis is defined as a temporary conclusion from the results of observations in order to produce a new theory.
4. Erwan Agus Purwanto and Dyah Ratih Sulistyastuti
Erwan Agus Purwanto and Ratih Sulawyastuti argued that a hypothesis is a temporary allegation of problems raised by researchers in conducting research whose existence is still weak. Because it is still weak and not necessarily true, testing is needed.
More simply, the meaning of the research hypothesis according to Dantes is an assumption that needs to be tested data. Then from testing through research will produce data. This data will be used as a reference for drawing conclusions, sometimes also generate new solutions and discoveries.
6. Fraenkel dan Wallen
Fraenkel and Wallen are more focused on meaning that the type of undirected hypothesis describes research conducted by researchers who do not make predictions, causing unclear directions and will affect the results of the research itself.
Talking about the undirected hypothesis actually includes an alternative hypothesis (Ha). Besides Ha, there is also a null hypothesis (Ho). There are two types of alternative hypotheses, namely directional hypotheses and non -directional hypotheses .
The undirected hypothesis is a hypothesis made by the researcher by explicitly formulating the problem and the researcher has also stated that the independent variable already has an influence on the dependent variable. The so-called undirected hypothesis is a hypothesis that has not been formulated explicitly, and the independent variables do not necessarily have an influence on the dependent variable.
7. Suharsini Arikunto
For those of you who have done research several times, you must be familiar with Suharsini Arikunto. He interpreted the hypothesis not much different from previous opinions. Broadly speaking, a hypothesis is a temporary answer to a problem being investigated by researchers. Until the research is complete, then the hypothesis can be proven through the data obtained and collected, whether it is appropriate or not.
Zikmund defines a hypothesis as a proposition (conjecture) that has not been proven. In other words, the alleged statement is still tentative (temporary). To explain it requires facts or phenomena (research studies) that allow answers to these propositions.
The definition of research hypothesis according to Sudjana is a temporary assumption (allegation) of something that is made. Generally, this assumption is made to explain something that requires confirmation or checking.
Sugiyono interprets the hypothesis as a temporary answer made based on the research problem formulation that has been determined by the researcher. Writing the formulation of the problem is packaged in the form of a question. The hypothesis is said to be temporary because the conjecture is based on theory, so hypothesis testing is needed.
Types of Hypotheses
In inferential research, especially in correlation and comparative research, hypotheses are classified into two, namely hypotheses without direction which are also called two-way hypotheses and unidirectional hypotheses, as explained below.
1. Directionless Hypothesis (Two Way)
A directionless hypothesis is a formulation (sentence) of a hypothesis that contains statements only regarding a relationship or only a difference, without explaining the direction of the relationship between the variables studied, for example, positive (+) or negative (-) directions. For example, the directionless hypothesis “There is a significant relationship between Learning Motivation and Student Achievement”. In this example, the direction of the relationship (whether towards a positive or negative relationship) between the variables of learning motivation and student achievement is not explained.
Another example, the hypothesis that reads “There are significant differences in student achievement based on learning motivation”. This hypothesis also does not include an explanation of which learning motivation has high learning achievement.
2. Unidirectional Hypothesis
The unidirectional hypothesis is generally structured as a statement indicating the direction of the relationship or difference of the two variables studied; direction reflects a positive or negative relationship. For example, the research hypothesis “The higher the student’s learning motivation, followed by the higher the student’s achievement”; indicates the direction of a positive relationship. Another example “The higher the self-concept, followed by the lower the student’s aggressiveness”; which indicates that there is a negative relationship.
How to Develop a Hypothesis
It should be understood that research hypothesis formulations do not “fall from the sky” or appear suddenly without being based on a theory or scientific study. The research hypothesis is not formulated simply by following the assumptions or assumptions of the researcher, even though the researcher’s assumptions can be a starting point in theoretical studies and predictions of future research results. So, the formulated hypothesis does not just follow the assumptions or assumptions of the researcher, but originates from the elaboration of the theoretical basis previously prepared.
The theory links the existence of the independent variable with the dependent variable. Therefore, theoretical analysis and relevant research findings serve to explain the problem and establish predictions about the answers to the research questions.
As stated by Azwar (1999), that in formulating a hypothesis, there are two ways. The first way is to read and review (review) the theories or concepts that discuss the research variables and their relationships. This method is often referred to as a deductive thinking process. The second way is to read and review the results or findings of previous studies that are relevant to the research problem.
This is known as an inductive thinking process. After reviewing the theories and research findings, researchers can formulate research hypotheses. The results of the theoretical study as well as the findings of the research results are an important provision (foundation) for researchers in preparing their hypotheses. Therefore, in general, the hypothesis is placed after the researcher examines the theories, concepts and research findings, namely at the end of chapter II of a research report.
The hypothesis must be tested for validity through statistical tests using appropriate analytical techniques. The hypothesis that has been prepared needs to be verified using advanced statistical analysis techniques. The choice of statistical analysis technique depends on several things, namely the type of research, research objectives and the type of data scale on each variable.
In the formulation of the hypothesis statistically expressed through symbols. There are two kinds of hypotheses, namely
the null hypothesis (Ho) and the alternative hypothesis (Ha), which are always written in pairs. If one is rejected, the other must be accepted, so that a firm decision can be made, that is, if H0 is rejected, Ha will be accepted. By pairing it, a firm decision can be made, which one is accepted and which one is rejected.
Below is an example of a statement that can be formulated as a statistical hypothesis:
- In an experimental study entitled “The effect of traditional learning models on students’ pro-social abilities”, the formulation of the statistical hypothesis is structured as follows: Ho : There is
no effect of traditional learning models on students’ pro-social abilities
Ha : There is an influence of traditional learning models on pro-social abilities -Social students
- In the experimental research entitled “Effectiveness of BK Services on Increasing Student Confidence”, the formulation of the statistical hypothesis is structured as follows:
Ho: BK services are not effective in increasing student self-confidence
Ha: BK services are effective in increasing student self-confidence
Proof of Research Hypothesis
In inferential research that has to test a hypothesis, including experimental research, proving a hypothesis is always related to the term significance. An understanding of the significance level is very important in the use of statistical methods to test hypotheses. This is due to the fact that inferential research conclusions are always based on statistical decisions, which cannot be supported by one hundred percent absolute confidence level.
In inferential research, researchers always use probability, namely the possibility of error in rejecting or accepting a hypothesis. In analyzes that use statistics, the significance level (sig) is often given the symbol p or the symbol alpha (α) expressed in a proportion or percentage, which means the magnitude of the probability of
According to the consensus of statisticians, the highest acceptable error probability is 0.05 or 5%; means that the chance of error is 5%, meaning an error of 5 out of 100 events. Conversely, this also means that the confidence level is 100-5 = 95% or 0.95. In social research, especially in the field of education, the significance level is generally measured from a p of 1%, or 5%.
When conducting research analysis, the researcher especially needs to read (interpret) the results of Sig (p), and then read the value (score) of r (correlation coefficient). Whereas in the different test study, after the researcher read the sig results, followed by the t score (t-test results), or F (Anova results), and the r square score (r2).
It should be emphasized again that the significance of the research results (probability of error) is referred to from the level of significance (p or sig) found. In the research analysis, the distribution of the probability of error (sig) is divided into three groups, namely:
- p <0.01, then the correlation or difference is stated to be very significant. Thus the hypothesis is accepted!
- p <0.050 (between 0.011 – 0.050), then the correlation or difference is stated to be significant. Thus the hypothesis is accepted!
- P > 0.05, then the correlation or difference is stated to be non-significant (not significant). Thus the hypothesis is rejected!
For example, experimental research on the Effect of Assignment Learning Models on Student Learning Motivation, which produces sig = 0.089, and the magnitude of r square is 0.061. This means that in this study there was no significant effect of the assignment learning model on student learning motivation. Thus, the hypothesis which reads “Assignment Learning Model influences Student Learning Motivation” is rejected. The effective contribution of the assignment learning model to the existence of student learning motivation appears to be low, namely only 6.1%.
There are differences in proving (testing) hypotheses in inferential research (including experimental research) and action research. Proving the hypothesis in inferential research always uses statistical tests, as described above. Acceptance or rejection of a hypothesis is assessed from the results of its significance score. If the significance score obtained is more than 0.050 then the research hypothesis is not significant or is rejected. Whereas in action research, hypothesis testing is assessed from the results of each action compared to the formulation of the research achievement indicators.
Thus, in action research researchers need to formulate indicators of achievement. If the results of the action have exceeded the achievement indicators, the research is proven to have succeeded in achieving its objectives.