P ROCEDIMIENTO PARA LA G ESTIÓN DE C OMUNICACIONES

The developed translated instruments were tested for their reliability and validity. The reliability assessment used internal consistency, spilt-half analysis, and test-retest, while the validity was evaluated against the criteria.

4.3.7.1 Pilot study

A small pilot study was undertaken to assess the clarity and understanding of the questions and instructions; it was also used to identify any procedural problems in the distribution and collection of the completed questionnaires. From the targeted population, the PHCC Wasat Abha, in the Asir Region, Saudi Arabia, was randomly selected out of 12 PHCCs in Abha city.

This PHCC (or pilot centre) was excluded from the main study to avoid compromising the validity of the results due to participant familiarity with the instruments. To promote the pilot study, posters were placed inside the PHCC to recruit participants. The questionnaire was distributed to 36 PHC professionals, along with a plain sheet of paper for the professionals to write their comments and feedback about the questionnaire. A total of 28 questionnaires were returned. One questionnaire was incomplete, thus providing 27 usable questionnaires. The effective response rate for the pilot study was 75 percent. The average time to complete the questionnaire was 20-25 minutes. The participants were encouraged to write comments about the questionnaire’s clarity and other feedback which they thought might be useful for the researcher.

The feedback was given in relation to question 9 in the demographic section; it was related to the classification of the PHCCs. The overriding comment was that the participants were familiar with the notion of a classification level for the PHCC. This classification system, implemented in 2010, appears to have been instigated without any PHCC staff orientation. Thus, the researcher changed the question to another question w, that is, one about the size of the PHCC, based on the number of staff, to explore whether the size of the PHC centre affected the satisfaction and organisational culture of PHC professional.

Interestingly, three participants commented, “it is good someone hears about our problems”, and that “these questions are talking about our situation and environment as you are living with us”.

Importantly, the participants indicated that they did not experience any difficulty with rating the scale. Further, the instructions appear to have been written clearly as the participants were able to follow them correctly. Their comments indicated that the questionnaire was acceptable and readable.

The data from the pilot study were analysed using the Statistical Package for Social Services (SPSS) for Windows (version 19). Testing the reliability was the first step or procedure used to evaluate the test/measurement; this test refers to the precision of the measurement and indicates the consistency of the measurement’s scoring procedure. Additionally, the spilt-half reliability test was performed to provide an estimation of the correlation (Thorndike, 2005)

4.3.7.2 Demographic characteristics of pilot sample

In terms of the demographic characteristics of the participants in the pilot study, over half were nurses (59.3%), one quarter were physicians (25.9%), and over one-tenth (14.8 %) were allied health professionals (see Table 4.5).

Table 4.5 Frequencies and percentages of respondents of Wasat Abha PHCC professionals

Profession Frequency Percent

Nurses 16 59.3

Physicians 7 25.9

Allied health professionals 4 14.8

Total 27 100

The length of experience of the PHC professionals in the pilot study ranged from one year to over seven years. Table 4.6 give an overview of frequency distribution for the participants’

years of experience. Over half the of the participants had six or more years of experience in a PHC setting, while 3.7% had three years’ and four years’ experience, respectively.

Table 4.6 Years of experience of respondents in pilot study in descending order of frequency and percentage

Years of Experience Frequency Percent

Above seven years 9 33.3

Six years 6 22.2

One year 4 14.8

Two years 3 11.1

Five years 3 11.1

Three years 1 3.7

Four years 1 3.7

Total 27 100

4.3.7.3 Reliability and correlation

4.3.7.3.1 Cronbach's alpha (α)

To ensure the stability of the instrument used for the scope of the study, a Cronbach’s alpha-coefficient was calculated for each question in the questionnaire, as well as overall, to determine the study’s internal consistency. The results of the analysis (Cronbach’s α = 0.93to 0.96) indicated consistency for all sections (see Table 4.7). These results confirm the stability of the instrument as being very high, which strengthens the reliability of the questionnaire. Hence, the measurements and instruments used in the pilot study were deemed as suitable for use in the main study.

Table 4.7 General alpha reliability statistics for each section of the questionnaire Sections N of Questions Cronbach's Alpha

Section B: OCI 24 0.93

Section C: LBDQ 40 0.96

Section D: MMSS 31 0.94

4.3.7.3.2 Split-half reliability

The split-half reliability test was another measure used to determine the consistency of the instrument; it splits the score in two halves, and compares one with the other. Then it estimates the correlation between the two total scores; if the test is consistent, it is most likely to be measuring the same construct. All sections of the instrument were found to be consistent with same half of the section. Table 4.8 shows the three instrument split-half reliability. Therefore, the consistency, reliability, and correlations were found to be high and confirmed the estimate reliability of the study’s questions and survey.

Table 4.8 Split-Half Reliability of all sections of the study

Sections SPSS Split No. of

Questions

Reliability Spearman

Section B: OCI 1st first half Split 12 0.80 0.94

2nd second half Split 12 0.91

Section C: LBDQ 1st first half Split 20 0.93 0.93

2nd second half Split 20 0.91

Section D: MMSS 1st first half Split 16 0.86 0.88

2nd second half Split 15 0.93 4.3.7.3.3 Correlation

The correlation matrix of the OCI, Section B (see Table 4.9) showed the presence of several significant correlations at the level of alpha = .05 and above. While, two questions (5 and 11) did not reach statistical significance, the result of Cronbach’s alpha coefficients did not indicate a weakness for these questions. Therefore, the researcher included these two questions in the main study.

Table 4.9 The correlation matrix of OCI, Section B

Item Corr. Item Corr. Item Corr.

Q1 0.51** Q9 0.71** Q17 0.58**

Q2 0.68** Q10 0.63** Q18 0.82**

Q3 0.51** Q11 0.13 Q19 0.71**

Q4 0.56** Q12 0.56** Q20 0.52**

Q5 0.14 Q13 0.81** Q21 0.69**

Q6 0.59** Q14 0.77** Q22 0.77**

Q7 0.76** Q15 0.63** Q23 0.76**

Q8 0.79* Q16 0.63** Q24 0.74**

Corr. Is (**) sig. at (0.01), Corr. Is (*) sig. at (0.05)

The correlation matrix of LBDQ, Section C of the instrument (see Table 4.10) shows the presence of several significant correlations of 35 items at the level of alpha = .05 and above.

While five items (7, 19, 23, 28, 33) showed a non-significant correlation, the Cronbach’s alpha coefficients results indicated no weakness in these items. Therefore, the researcher included these questions in the main study.

Table 4.10 The correlation matrix of LBDQ, Section C

Item Corr. Item Corr. Item Corr. Item Corr.

Q1 0.64** Q11 0.64** Q21 0.74** Q31 0.70**

Q2 0.59** Q12 0.72** Q22 0.79** Q32 0.54**

Q3 0.77** Q13 0.76** Q23 0.30 Q33 0.27

Q4 0.64** Q14 0.57** Q24 0.40* Q34 0.76**

Q5 0.65** Q15 0.65** Q25 0.72** Q35 0.82**

Q6 0.74** Q16 0.5** Q26 0.51** Q36 0.55**

Q7 0.38 Q17 0.74** Q27 0.69** Q37 0.55**

Q8 0.75** Q18 0.78** Q28 0.07 Q38 0.60**

Q9 0.71** Q19 0.13 Q29 0.82** Q39 0.74**

Q10 0.64** Q20 0.73** Q30 0.73** Q40 0.78**

Corr. Is (**) sig. at (0.01), Corr. Is (*) sig. at (0.05)

The correlation matrix of the MMSS (Section D) instrument (see Table 4.11) revealed the presence of several significant correlations for 28 items at the level of p <0.05 (*) and above.

While three items (2, 8, and 14) were not statically significant, the alpha Cronbach coefficients did not indicate any weakness in the questions. The researcher therefore included these questions in the main study.

Table 4.11 The correlation matrix of MMSS, Section D

Item Corr. Item Corr. Item Corr.

Q1 0.58** Q11 0.80** Q21 0.73**

Q2 0.29 Q12 0.66** Q22 0.87**

Q3 0.60** Q13 0.56** Q23 0.79**

Q4 0.46* Q14 0.19 Q24 0.68**

Q5 0.53** Q15 0.60** Q25 0.64**

Q6 0.61** Q16 0.41* Q26 0.74**

Q7 0.61** Q17 0.59** Q27 0.73**

Q8 0.33 Q18 0.57** Q28 0.71**

Q9 0.64** Q19 0.62** Q29 0.58**

Q10 0.50** Q20 0.64** Q30 0.64**

Q31 0.77**

Corr. Is (**) sig. at (0.01), Corr. Is (*) sig. at (0.05)

From the results of the correlations it is clear that most questions showed a high level of significance, with the most abstract level being (0.01) (**), as shown in Tables 4.9, 4.10, and 4.11. Therefore, the results confirm that the research tool has the appropriate level of psychometric properties and, therefore, is deemed appropriate for use in the current study.

The internal consistency coefficients (correlation) between the total score for each axis and the total score for the instrument of the study ranged from 0.87 ** to 0.91 **

(see Table 4.12), all of which function at the level of significance of (0.01) (**). The result emphasises the strength of the study questionnaire.

Table 4.12 The internal consistency coefficients of each section and the total score for the study survey

Sections No. of Items Correlation with Grand Total

B (OCI) 24 0.89**

C (LBDQ) 40 0.91**

D (MMSS) 31 0.87**

Corr. Is (**) sig. at (0.01)

These results establish the overall stability of the instrument, given its high level of internal consistency and reliability. Therefore, the instrument was deemed suitable for the final study.

Further, it was expected to give the same results in any similar environment, that is, in terms of time and place.

4.3.7.4 Criterion validity

Criterion validity is the degree to which the results predict what it is being measured and that it is theoretically similar (American Educational Research, 1999). Hence the assumption underlying the current study was that: “Organisational cultures were significantly (positively) correlated with leadership behaviour and job satisfaction, and leadership behaviour was significantly (positively) correlated with job satisfaction” (Tsai, 2011, p. 89). Hence, it was expected that the OCI and LBDQ would be a predictor on the job satisfaction scale. The higher the correlation, the better the test (Thorndike, 2005).