Transform your insight generation process
Use our in-depth online survey guide to create an actionable feedback collection survey process.
SHARE THE ARTICLE ON
Assessment of the quality of research helps in gauging how the research was meaningful in contributing to the effective understanding of the topic of study. Two of metrics that are relied upon to achieve this assessment are validity and reliability.
Validity refers to the accuracy with which a thing is measured. The validity is a measure of authenticity , in the sense that it projects the true and accurate picture without any window dressing or distortion of any kind.
For example: A study that is intended to gather the market sentiment towards a product that an MNC is going to launch, will probably try to gather maximum opinions from the people belonging to their target market . This research will be considered valid if the changes made in the product based on the suggestions and point of views given by the sample population during the study results in an improvement in the company’s market position in terms of customer acquisition , increase in market share or increase in marginal revenue.
Reliability refers to the usefulness of data in fueling informed decisions . When the same results are obtained consistently for the experiment under consideration, it becomes an indication of the reliability.
Reliability is a qualitative metric that guides the direction of follow up. The more reliable a data is , the better it will be able to direct the flow of intelligent decision making.
For example : A product feedback website projects the initial responses to a newly launched product in good light. The customers who have provided the feedback are seen mentioning the product as being highly effective and pocket friendly. On the other hand, customer review blogs and other websites highlight major drawbacks that come with the product. The product is said to be faulty , sub-par and the customer service offered by the brand that launched it is claimed to be very poor.
The inconsistency in the customer feedback regarding the product in the above example indicates that the product feedback website initially visited does not reflect a true picture of the customer opinions and thus , is not a reliable source for a potential customer to make decisions about their product purchases.
Validity and reliability share a complementary relationship.
This means that the existence of accuracy implies reliability. If a data is deemed accurate such that it reflects the intended metric in the most precise manner , it naturally becomes reliable and so can be used as input to strategize and make important follow-up decisions.This even reduces the risk factor associated with taking misdirected decisions which can land organizations in a predicament.
The reverse is also true : if a data is reliable and has been used by other entities for their working , it indicates that the data measured is correct and potrays a true picture. The underlined idea here is that if other organizations trust the authenticity of the data to use it for achieving their own objectives , the data is correct and its validity is high.
The test- retest mechanism runs on the idea that a relliable data remains consistent over time. This implies that the results of an experiement or an activity is reliable, provided that it repeats itself when the same activity is conducted at a different. The continuous delivery of uniform results provides the assurance that the results so obtained are correct and can be depended upon for efficient and effective decision making.
For example : Movie and show likings remain consistent over time. So once a person has viewed a particular TV show or a movie, they generate an opinion regarding it which remains constant irrespective of other genres that they watch. If an OTT platform’s market research shows that people watched and rated a movie highly but changed their opinion to rating it drastically bad when asked subsequently, it is reflective of a non-uniform pattern and so is not considered reliable. According to Test-Retest methodology , the market research of the OTT platform could only be considered reliable if the results of the subsequent research would have also shown people’s views about the movie as being a good watch, provided the research is carried out using the same target group.
Results obtained on one aspect must be correlated to results from another aspects if all of them assess the same underlying idea or principle.
For example: If the prices of an LED TV have gone up because of irregular demand and excess production, then the studies gathering demand of customers for that LED TV and production figures for that time period must be in line with this increase in price for the results to be reliable. If the production figures show that the manufactuing of the TV has been reduced to meet the reduced demand, then there will be a lack of internal consistency due to which the increase in production cannot be identified as a plausible cause of increase in prices.
The different aspects must be correlated and indicative of a common conclusion that can be effectively relied upon for easy decision making.
Interrater reliability states that different people trying to measure the same thing must come up with uniform conclusions. This is based on the concept that different people judge similar concepts in aa similar manner , provided that they’re judgement and approach is correct and standardized. This is confirmation technique wherein the observation or judgement of one person gets confirmed through the judgement of more people who verify the initial claim. If the subsequent judgement after the initial one does not outline the same idea, it means that the judgement is faulty and inreliable.
For example : The claim that a students is an excellent badminton player can only be relied upon if other people with a comprehensive understanding of the sport and its rules , back up the initial claim by providing their positive opinion about the student’s ability to play badminton. If the assessment of different people is different , then the conclusion that the students is a good badminton player is questionable.
Content validity is the extent to which a concept is covered thoroughly . This involves breaking down the concept which is to be measured into essential aspects that need to be studied for a complete understanding. Content validity ensures that flawed results are avoided as a result of incomplete study.
For example: An assessment of the level of confidence in an individual involves gauging tone, manner of speaking and body langauge. The comprehensive overview of a person’s confidence levels must include and analyse individuals on all the three bases. Covering these areas can help combine results obtained from individual aspects into an accurate conclusion. Failing to cover even one of these aspects can make the results skew in the opposite direction leading to judgements that not precise and correct.
Criterion validity measures how the results of a research correlate with other relevant outcomes . This shows the authencity of a result , judgement or opinion in making effective predictions and in delivering a standard result across related studies.
For example : The validity of an exit poll conducted by a news channel during elections will be measured by how effectively the exit poll data has been used to predict the outcomes of the election. If the predicted winner of the election is in line with the actual declared result, then the validity of the exit poll is very high and the partcular news channel that conducted the poll becomes a good source of gaining accurate political updates.
Unlike criterion validity , discriminant validity assesses the relationship between the concept under study and other concepts which have no relevance to it. This is to ensure that the results do not cover unnecessary subjects that have no value addition to the topic and can adversely affect end results.
For example : Transactional feedback that is provided after every customer interaction is a great example for studying discriminant validity. Customers are asked to rate their entire journey with a brand on a scale of 1-10 after they make a purchase. A customer rates their experience as 9/10, meaning that they find the brand and its products highly suitable for their needs and their experience with the brand has been pleasant overall. Next week , the same customer answers the same question ,after having experienced poor service at the outlet, by giving 3/10 .
The variation on the ratings given by the customer within such a short interval of time clearly shows that customer has used their recent experience of purchase as the basis of answering questions related to their entire customer journey instead of keeping the overall journey in mind. This makes their review , an improper representation of holistic customer experiences and so is highly innacurate.
In order to improve validity , focus on :
For increasing reliability , pay attention to :