research support services

CHAPTER THREE

METHODOLOGY

3.1 Introduction

The study aimed at investigating the effect of Information systems on performance of government agencies in Uganda. This chapter presents the research methods that were used to carry out the study. It covered the research design, area of study, target population, sample design, sample size, research instrument, measurement of variables, data collection procedure, data analysis and ethical considerations of the study.

3.2 Research Design

The study used a cross-sectional research design. This design was used because the researcher was able to collect data from across all corners of the population of interest in the study. This design was adopted because it was cheap in terms of time and cost as observed by (Mugenda, 2003, Sekaran, 2004). Both quantitative and qualitative research approaches for data collection and analysis was used in this study.  According to Mugenda (2003), the two approaches supplement each other, and they help to reduce bias in each approach.  Amin (2005) argues that triangulation enables the researcher to have a deeper analysis using the inductive and deductive approaches through qualitative and quantitative perspectives, which enable the researcher to analyse data from all angles, and give a more concrete and realistic description of the findings. Qualitative approach helps in interpreting peoples’ opinions, perceptions, and attitudes to give a deeper understanding into the problem under investigation.

3.3 Study Population

The study population was generated from URA and comprised of 239 employees, 1 Executive

Director,12 management staff, 40 Division Heads, 5 Regional Heads, and 181 staff members at URA whose duties influencedthe effect of Information systems on performance of government agencies.

3.4 Determination of the sample size

Using Krejcie and Morgan’s (1970) table for sample size determination approach, a sample size of 181 was determined from the total population of 239 employees and 100 prominent tax payers.

Table 3.1: Showing Population and Sample size of respondents

Source: URA Employee List, (2013)

3.5 Sampling techniques and procedure

This study used both probability and non-probability sampling techniques. In probability sampling, simple random sampling was used, while purposive sampling was used for nonprobability sampling.

3.5.1 Simple random sampling

In this sampling technique every element/member in the population had equal chances of being selected to participate in the study. This technique was used because it avoids bias, and is easy to use. Neuman (2006). It was used to select staff members and tax payers.

3.5.2 Purposive sampling

Purposive sampling strategy was used in this study because it enabled the researcher to use cases in the population that have the required information, as argued by (Mugenda 2003; Neuman 2006).  In purposive sampling the researcher approached those respondents that were more knowledgeable about the problem under investigation thus enabling her not to waste time and resources on respondents with less/no information on the problem. It was used to select Executive Director, Managers, Division Heads and Regional Heads.

3.6 Data collection methods

The section presents data collection methods which include; questionnaire survey, interview and documentary review.

3.6.1Questionnaire Survey

Questionnaire Survey method was used to obtain the opinion of the respondents regarding the topic under study. According to (Onen & onen, 2013) questionnaires are important in research because the respondents are given time to think and they don’t feel intimidated. Questionnaire gives the respondents ample time to respond to the questions when ready and they can be kept for future references. This method was deployed to capture information from Staff Members,

Regional Heads, Division Heads and tax payers.

3.6.2 Interview

Interviews are an effective qualitative method for getting people to talk about their feelings, opinions and experiences. They are also an opportunity for us to gain insight into how people interpret Information systems on the performance of government agencies. The views of the respondents were a personal reflection of their personal experience relating to the study topic. Open ended questions allowed ease of expression and capture of vast information from study participants.  This method was deployed to capture information from the Executive Director and managers. Appendix III presents the interview guide.

3.6.3 Documentary review

The researcher analysed the documents and publications related to the study topic. Documents that were reviewed include; URA reports, Journals, research publications, magazines, text books and Newspapers.

3.7 Data collection instruments

For each deployed data collection method, there is a corresponding data collection instrument that was used. The study used Questionnaire Guides, Interview Guide and Document review checklist as described in the sub-sections below.

3.7.1 Self-administered Questionnaire

The questionnaire was designed in a manner that motivates respondents with simple structured questions with the option of providing any addition information to the structured questionnaire as an option to obtain relevant data from them. The questionnaire was structured with both close-ended and open-ended questions. It had a Likert scale 1-5 indicating the level of a respondents’ agreement or disagreement, where 1 represents Strongly Disagree and 5 Strongly agree. The questionnaire is attached in Appendix I and II.

3.7.2 Interview Guide

Unstructured interview guide was designed and used by the researcher to collect qualitative data from key informants in Uganda Revenue Authority. It had key items/questions that were asked to key respondents and then it was filled by the researcher while conducting a face to face interview.  This instrument according to Sekaran (2004) enabled the researcher to collect rich and detailed data, get more clarifications, and it enabled her to establish rapport and motivate respondents to answer questions.  Yuko & Onen (2009) argue that this instrument enables the researcher to collect data that can’t be written, and to capture meaning beyond words, and it yields a high response rate. The researcher used the interview guide as seen in (appendix III) to supplement data got from questionnaire and get more clarification on variables under the study especially Information systems on the performance in Uganda Revenue Authority.

3.7.3 Document Review Checklist

The researcher used this instrument in order to capture secondary data and first-hand   information relevant to the study. These documents helped the researcher by revealing the level of Information systems and performance of URA. This was achieved through a review of the analysis reports, journals and newspapers.

3.8 Data quality control of instruments

The data collection tools were pre-tested on a smaller number of respondents from each category of the population to ensure that the questions were accurate.

3.8.1 Validity

Validity is defined as the extent to which results can be accurately interpreted and generalized to other populations (Oso & Onen, 2008). While Borg & Gall, 1989 as cited in Onyinkwa,

(2013) validity is defined as the degree to which results obtained by the research instrument correctly represented to the phenomenon understudy. Mugenda & Mugenda, (1999) defines it as the accuracy and meaningfulness of inferences which are based on the research results.  The formula for Content Validity Index was;

CVI = ^𝑛

𝑁

Where CVI = content validity  n= number of items indicated relevant.

N = total no. of items in the instrument.

In this study, validity was achieved by establishing content validity. The researcher achieved content validity by using the experts to assess the validity of the research instrument. The experts especially research supervisors and consultants from UMI were given data collection tools to assess whether the items in the instruments were valid in relation to research topic, objectives, and questions. From the instruments, they declared some items valid and others invalid. Those declared invalid were dropped, others adjusted, while the valid ones were maintained.  Then content validity index (CVI) was computed by dividing the number of items declared valid by total number of items/questions in the data collection instrument.

Table 3.2 Shows the content validity index (CVI) of the research instruments

Source: primary data

Therefore, CVI =0.763(76.3%).

From table 3.2, CVI was 0.82 (82%), and this was very good.  According to Waner (2005), as cited in Barifaijo, Basheka and Oonyu (2010), if the CVI is greater than 0.7, then the instrument is said to have a high content validity. The researcher analysed the data collected and where need arose, the instruments were re-adjusted and re-designed to improve reliability and validity. To improve face validity a pilot study was carried out at URA.

3.8.2 Reliability

According to Mugenda and Mugenda, (2003) reliability is the measure of the extent to which research instruments are able to provide the same results upon being tested repeatedly.

Crobach’s coefficient alpha (a) as recommended by Amin, (2005, P.302) was used to test the reliability of the research instrument. The instrument is deemed reliable if reliability of 0.7 and above is obtained and therefore, it was adopted for use in the data collection.

Formula for reliability is

∝= 𝐾      (∈𝑆𝐷22𝐼)

𝐾−1                  𝑆𝐷 𝑡

Where ∝ = alpha reliability co efficiency.

K=Number of items included in the questionnaire

∈ 𝑆𝐷²𝐼 = sum of variance of individual items

𝑆𝐷²𝑡 = variance of all items in the instrument.

To ensure credibility and trust worthiness of qualitative data the researcher ensured that only the officials who were employees of URA were interviewed.

The coefficient ranges between a=0.00 for no reliability, a =1.00 for perfect reliability. The closer alpha gets to 1.0 the better. If the study findings result to Cronbanch’s Alpha of 0.7 and above, this signified that the research instrument was good enough for the study. According to Amin (2005), all the measurements in the instrument that show adequate levels of internal consistency of Cronbach’s alpha of 0.77 and above are accepted as reliable.

Table 3.3: Shows Reliability of research instruments

Source: primary SPSS data

Overall reliability= 4.621/6 =0.77 (77%)

The table 3.3 shows reliability of instruments on different variable, with an average Alpha of 0.77(77%), and this was good enough for the study according to Mugenda &Mugenda (1999) and Amin, (2003).

3.9 Procedure of data collection

The researcher obtained an introductory letter from Uganda management institute to seek permission and enable easy access of information by the researcher from URA. After the permission was granted from

URA, the researcher went ahead and administered questionnaires and interviewed selected respondents. However, the consent of the respondents was sought before being given questionnaire and the respondents were informed that the study was strictly for academics.

3.10 Data analysis

The study used both quantitative and qualitative data analysis methods

3.10.1 Quantitative Data Analysis

Data processing was done by entering the data into a statistics package for social sciences (SPSS) version 24.0 in line with the research questions. Data analysis was done by also using this statistics package for social sciences (SPSS) to formulate frequency tables where the percentages, frequency, mean, variance and standard deviation were obtained.

The quantitative analysis process included; editing, classification, coding and presentation. Data was summarized in frequency tables; percentage and it was analysed with the use of statistical package for social scientist (SPSS). Quantitative data was collected through structured questionnaires and it was entered into a computer, tabulated and analysed.

Spearman’s correlation coefficient and regression analysis is recommended by Amin (2005, P.378) was used during data analysis in order to test the strength, degree and direction of the effect of Information systems on performance of government agencies. The formula was used for this study because it was compatible with SPSS program in addition to being appreciated in analyzing data under which it was arranged.

3.10.2 Qualitative Analysis

Qualitative data was analyzed using content analysis. It involved gathering and analyzing data based on the content, where by the raw data collected from the field was read through to enable the researcher to get familiar with the data. At this process, the study used noted cards to organize the available data to accelerate further analysis. Data was then evaluated and analyzed to determine its accuracy, credibility, usefulness and consistency which aided acceptance of the study.

3.11Measurements of variables

A five-point Likert ordinal scales ranging from; strongly agree which was assigned 5, 4 agree, Not Sure assigned 3, Disagree allocated 2 and strongly disagree allotted 1 to obtain responses on the variables. The Likert ordinal scale has been used by numerous scholars who have conducted similar studies such as Bowling, (1997).

The structured questions were measured using the following variables;

• Information systems software
• Information systems infrastructure
• User knowledge and skills

3.12Ethical considerations

The researcher ensured that before giving questionnaires to the respondents their consent was sought and when they accepted to participate in the study, they were given questionnaires.

Confidentiality of the respondents’ information was assured and the researcher also informed them that the study was strictly for academic purposes and therefore, they should not fear giving information.

Only respondents who were selected were given questionnaires and only those meant to be interviewed were actually interviewed.