Dv Wk3 Kirk (2019) notes the importance of formulating your brief. What does he mean by this? Please expand this thought by noting how you would create a vision for your work. Note any real-world examples to expand upon this thought. Learning objectives
Participants will be able to:
Understand different ways of summarizing data
Choose the right table/graph for the right data and audience
Ensure that graphics are self-explanatory
Create graphs and tables that are attractive
By the end of this session, participants should be able to: [READ BULLETS]
Data Presentation, Interpretation and Use
By the end of this session, participants should be able to: [READ BULLETS]
Do you present yourself like this?
Do you present yourself like this? [HAVE AUDIENCE ANSWER QUESTION.]
Why would you not present yourself like this? Do you think this man is taken seriously? What do you think would happen if he tried to speak to someone in the Ministry of Health about some information related to a BCC campaign? Would he even be let in?
So, if you know that you would not be taken seriously if you presented yourself like this, then . . .
So why would you present your data like this?
Why would you present your data like this? Would most people be able to get the message from this data if it was presented in this STATA output? [ALLOW COMMENTS]
No, it is too busy and it is difficult to interpret.
The way you present your data can greatly affect how usable the data will be.
And why would you present your data like this? Can anyone tell me what some problems may be with this chart?
No axis labels
The colors are difficult to read. (You should never put a dark color on a dark background.)
The green color is too bright.
This is Better!
Use of ITNs in Zambia
What is improved in this slide compared to the last one? (other than the data points themselves)
The colors are easy to read.
% of children under 5 who slept under an ITN last night % of children under 5 who slept under an ITN last night % of children under 5 who slept under an ITN last night
% of women 15-49 who slept under an ITN last night % of women 15-49 who slept under an ITN last night % of women 15-49 who slept under an ITN last night
2001-02 DHS 2007 DHS Column1
% of children under 5 who slept under an ITN last night 7.3 28.5
% of women 15-49 who slept under an ITN last night 8 28.2
To resize chart data range, drag lower right corner of range.
Regardless what communication formats you use, the information should be presented in a clear, concise way with key findings and recommendation that are actionable.
For all communication formats it is important to ensure that there is:
Font, Colors, Punctuation, Terminology, Line/ Paragraph Spacing
An appropriate amount of information
Less is more
Appropriate content and format for audience
Scientific community, Journalist, Politicians
An appropriate amount of information will be determined by your audience and format.
Policymakers may do better with direct and concise summaries of key points, whereas the scientific community will want more detail.
On a PowerPoint slide, try to limit to six lines with no more than six words per line, balance text with graphics, and make sure that there are not too many slides.
One way to ensure that you create consistent materials is to decide on a template for the document/presentation/graph, etc., before you produce it. You can then give these guidelines to the different people involved in the process, and then only have to do minor formatting at the end.
Simplest way to summarize data
Data is presented as absolute numbers or percentages
Charts and graphs
Visual representation of data
Usually data is presented using percentages
The two main ways of summarizing data are by using tables and charts or graphs.
A table is the simplest way of summarizing a set of observations. A table has rows and columns containing data which can be in the form of absolute numbers or percentages, or both.
Graphs are pictorial representations of numerical data and should be designed so that they convey at a single look the general patterns of the data. Generally, the data in a table is in the form of percentages. Although they are easier to read than tables, they provide less detail. The loss of detail may be replaced by a better understanding of the data.
Tables and graphs are used to
Convey a message;
Stimulate thinking; and
Portray trends, relationships, and comparisons.
The most informative graphs are simple and self-explanatory.
Tables can be good for side-by-side comparisons, but can lack visual impact when used on a slide in a presentation.
Points to remember
Ensure graphic has a title
Label the components of your graphic
Indicate source of data with date
Provide number of observations (n=xx) as a reference point
Add footnote if more information is needed
To make the graphic as self explanatory as possible there are several things to include:
Every table or graph should have a title or heading
The x- and y-axes of a graph should be labeled, include value labels such as a percentage sign, include a legend
Cite the source of your data and put the date when the data was collected or published
Provide the sample size or the number of people to which the graph is referring
Include a footnote if the graphic isn’t self-explanatory
These points will pre-empt questions and explain the data. In the next several slides, we’ll see examples of these points.
Tips for Presenting Data in PowerPoint
All text should be readable
Use sans serif fonts
Gill Sans (sans serif)
Times New Roman (serif)
Use graphs or charts, not tables
Keep slides simple
Limit animations and special effects
Use high contrast text and backgrounds
Rikki Welch (RSW) – edit
All text should be readable. Try to avoid having text in less than 25 point font. There are exceptions, of course (especially when creating and using graphs and charts), but try to make sure that everything is readable from the back of the room.
Use no more than 2 typefaces per presentation. In PowerPoint slides, a sans serif typeface can be more readable than a serif typeface.
Nothing in your slides should be superfluous (no extra doodads for decoration).
Limit the use of animations and other special effects. Use them sparingly, if at all.
Ideally, there should be no more than 6 lines per slide, with six words per line.
Resist the urge to add too many slides.
A light background with dark text (such as this one) will show up better a light-filled room than a slide with a dark background and light text.
Choosing a Title
A title should express
A title should most of the time express who, what, when, and where.
Tables: Frequency distribution
Year Number of cases
2000 4 216 531
2001 3 262 931
2002 3 319 339
2003 5 338 008
2004 7 545 541
2005 9 181 224
2006 8 926 058
2007 9 610 691
Frequency distribution is a set of classes or categories along with numerical counts that correspond to each one such as number cases in a given year.
What should be added to this table to provide the reader with more information?
Better labels-What type of cases? Malaria cases
Source of text on tables and graphs: Pagano M and Gavreau K. Principles of Biostatistics. 1993.
Percent contribution of reported malaria cases by year between 2000 and 2007, Kenya
Source: WHO, World Malaria Report 2009
Tables: Relative frequency
Year Number of malaria cases (n) Relative frequency (%)
2000 4 216 531 8
2001 3 262 931 6
2002 3 319 339 7
2003 5 338 008 10
2004 7 545 541 15
2005 9 181 224 18
2006 8 926 058 17
2007 9 610 691 19
Total 51 400 323 100.0
In this table, we already had the total number of observations (or n) in the second column but we added a title and the source of the data. Note that this table includes both a title and a reference. The citation is one area where it is acceptable to have typeface that is fairly small in relation to the rest of the text. You do want to have the citation on the slide so that people can know where the data is from if they want that information, but the citation is not the most important part of the slide. You want to draw attention to the data, not the citation itself.
We also added relative frequencies to this table. Relative frequency is the percentage of the total number of observations that appear in that interval. It is computed by dividing the number of values within an interval by the total number of values in the table then multiplying by 100. It is the same as computing a percentage for the interval.
To analyze this table, we should look at the relative frequencies. What do they tell us? There is an increasing trend in the number of reported malaria cases and in the relative frequency of cases.
Does this mean that there is an increase in malaria cases? What would this say about our programs?
It is important to take into account what we already know when interpreting these data. We know that since 2000 there has been an increased effort towards malaria control. During this time period, the quality of treatment has improved and the quality of routine information systems has improved.
When taking this knowledge into account how would we interpret these data?
From 2000-2007, the number of reported malaria cases increased. This may not reflect an actual increase in cases, but an increase in care seeking and reporting. Due to improved case outcomes seen after the introduction of ACTs in Kenya in 2004, individuals with fever began to seek care at formal medical facilities at higher rates. Furthermore, the routine information system improved during this period of time and thus reported more complete information.
Source of text on tables and graphs: Pagano M and Gavreau K. Principles of Biostatistics. 1993.
Use the right type of graphic
Charts and graphs
Bar chart: comparisons, categories of data
Histogram: represents relative frequency of continuous data
Line graph: display trends over time, continuous data (ex. cases per month)
Pie chart: show percentages or proportional share
We’re going to review the most commonly used charts and graphs in Excel/PowerPoint. Later we’ll have you use data to create your own graphics which may go beyond those presented here.
Bar charts are used to compare data across categories.
A histogram looks similar to a bar chart but is a statistical graph that represents the frequency of values of a quantity by vertical rectangles of varying heights and widths. The width of the rectangles is in proportion to the class interval under consideration, and their areas represent the relative frequency of the phenomenon in question A histogram is a histogram, not just because the bars touch. In the bar graph bars in a bar graph can touch if you want them to … but they don’t have to. Touching bars in a bar graph doesn’t mean anything.
In a histogram, however, the bars must touch. This is because the data elements we are recording are numbers that are grouped, and form a continuous range from left to right. There are no gaps in the numbers along the bottom axis. This is what makes a histogram.
Line graphs display trends over time, continuous data (ex. cases per month)
Pie charts show percentages or the contribution of each value to a total. When there are more than 4 categories it is best to go to a bar chart so that it is readible
In this bar chart we’re comparing the categories of data which are any net or ITN.
What should be added to this chart to provide the reader with more information?
Add a title and data labels. You could also add the source of the data but it isn’t necessary if all of your tables and graphs are derived from the same source/dataset.
On the next slide we see how the graph has been improved and is now self-explanatory.
Source: Quarterly Country Summaries, 2008
Note that this chart has a title, axis labels , data labels, and a source. It is best if you limit the bars to 4-8 to keep it readable, especially if it is to be used in a PowerPoint presentation.
Stacked bar chart
% Children <5 with Fever who Took Specific Antimalarial, 2007-2008 Speaker notes A stacked bar chart is often used to compare multiple values when the values on the chart represent durations or portions of an incomplete whole, such as the percentage of children taking each type of medication for fever when not all children received medication at all. * Histogram Speaker notes This is a histogram. At first glance, histograms look a lot like bar charts. Both are made up of columns and plotted on a graph. However, there are some key differences. The major difference is in the type of data presented on the x (horizontal) axis. With bar charts, each column represents a group defined by a categorical variable. This variable could be types of sports, different football teams, health facilities, or provinces. These are all categories. A histogram presents quantitative variables; the groups on the chart are always made up of numbers or something that could be turned into numbers. This could be age, height, weight, the number of minutes women wait in a queue, years, or months of the year. These groupings are sometimes called “bins.” The bin label can be a single value or a range of values. For example, you could split out the time spent waiting in line by the minute (5 minutes, 6 minutes, 7 minutes) or you could split it into chunks (less than 5 minutes, 6-10 minutes, 11-15 minutes). * Bar Chart v. Histogram * Data fabricated for illustration Speaker notes The columns in a typical bar chart can be arranged however you want to arrange them, alphabetically, by height, or the order in which you received the data—it doesn’t really matter. No matter which column comes first in this presentation, the idea presented does not change. * Chart1 22 15 29 Northwestern Copperbelt Central Average Clinic Wait Time in Minutes, by Province, 2012 Sheet1 Northwestern Copperbelt Central 22 15 29 To resize chart data range, drag lower right corner of range. Bar Chart v. Histogram (cont.) * Data fabricated for illustration Rikki Welch (RSW) - edit Speaker notes The order of the columns in a histogram is very specific, and the columns cannot be rearranged. The columns are arranged from low to high. A bar chart does not have a “high” end and a “low” end. A histogram does. You can see on this chart that the data is “skewed” toward the high end. It would NOT make sense to rearrange the columns on this chart. * Chart1 0-10 11-20 21-30 31-40 Over 40 Series 1 Average Waiting Time, in Minutes Percent of Total Clinic Waiting time in Eastern Province, 2012 2 3 10 30 55 Sheet1 Series 1 0-10 2 11-20 3 21-30 10 31-40 30 Over 40 55 To resize chart data range, drag lower right corner of range. Population Pyramid: Country Z, 2008 Speaker notes This is a population pyramid. It is basically two histograms presented side by side. On the right you can see males and on the left you see females. The bins shown are five-year age categories. Population pyramids are useful for presenting descriptive data about your population of interest or study population. On your disc, you will find a template for producing a population pyramid. All that you need is the data on age and sex and this excel worksheet will automatically produce a pyramid. * Line graph *Includes doctors and nurses. Number of Clinicians* Working in Each Clinic During Years 1-4, Country Y Speaker notes A line graph should be used to display trends over time and is particularly useful when there are many datapoints. In this case we have 4 datapoints for each clinic. By adding a label to the y-axis, a title and a footnote. In some settings, clinicians may only mean doctors but to be clear the footnote let’s the reader know that in this case we are referring to both doctors and nurses. * Chart1 Year 1 Year 1 Year 1 Year 2 Year 2 Year 2 Year 3 Year 3 Year 3 Year 4 Year 4 Year 4 Clinic 1 Clinic 2 Clinic 3 Number of clinicians 4.3 2.4 2 2.5 4.4 2 3.5 1.8 3 4.5 2.8 5 Sheet1 Clinic 1 Clinic 2 Clinic 3 Year 1 4.3 2.4 2 Year 2 2.5 4.4 2 Year 3 3.5 1.8 3 Year 4 4.5 2.8 5 To resize chart data range, drag lower right corner of range. Caution: Line Graph Number of Clinicians* Working in Each Clinic During Years 1-4, Country Y *Includes doctors and nurses. Speaker notes What is wrong with this line graph? If you look closely you can see that the X axis should be years, but instead it is clinics. Make sure that the right data is always charted on the axes, or else you may end up with a graph that cannot be interpreted like this one. * Chart1 Clinic 1 Clinic 1 Clinic 1 Clinic 1 Clinic 2 Clinic 2 Clinic 2 Clinic 2 Clinic 3 Clinic 3 Clinic 3 Clinic 3 Year 1 Year 2 Year 3 Year 4 Number of clinicians 4.3 2.5 3.5 4.5 2.4 4.4 1.8 2.8 2 2 3 5 Sheet1 Clinic 1 Clinic 2 Clinic 3 Year 1 4.3 2.4 2 Year 2 2.5 4.4 2 Year 3 3.5 1.8 3 Year 4 4.5 2.8 5 To resize chart data range, drag lower right corner of range. Pie chart Speaker notes A pie chart displays the contribution of each value to a total. In this chart, the values always add up to 100. What should be added to this chart to provide the reader with more information? What should be changed about this chart to make it more readible? POSSIBLE ANSWERS The color scheme, which is currently too bright The title should be more specific and indicate whether these are numbers or percentages. If these are percentages, that should be listed on the data and the n, or number of cases should be indicated to provide context. * Chart1 1st Qtr 2nd Qtr 3rd Qtr 4th Qtr Females Malaria Cases 59 23 10 8 Sheet1 Females 1st Qtr 59 2nd Qtr 23 3rd Qtr 10 4th Qtr 8 To resize chart data range, drag lower right corner of range. Pie chart N=257 Percentage of all confirmed malaria cases treated by quarter, Country X, 2011 Speaker notes A pie chart displays the contribution of each value to a total. In this case we used the chart to show contribution of each quarter to the entire year. For example, the first quarter contributed the largest the percentage of enrolled patients. To improve the understanding of the pie chart, we’ve added a more descriptive title and added value labels. On the previous chart, we couldn’t tell if the values are numbers or percentages. Adding the sample size let’s us know the total number of observations. For example It is also important to have charts that are attractive, easy to look at and easy to read. The chart on the previous page was so colorful that it was distracting, the colors were so bright that it was hard to look at the chart, let alone read it. While these colors are not the most interesting, they let the reader focus on the chart. The last chart was an exaggeration, but be sure to make sure that you do not make the same mistake on a smaller level. Limit the slices to 4-6. For extra pizzazz, contrast the most important slice either with color or by exploding the slice. * Chart1 1st Qtr 2nd Qtr 3rd Qtr 4th Qtr Females 0.59 0.23 0.1 0.08 Sheet1 Females 1st Qtr 59% 2nd Qtr 23% 3rd Qtr 10% 4th Qtr 8% To resize chart data range, drag lower right corner of range. How should you present… Prevalence of malaria in 3 countries over a 30 year period? Data comparing prevalence of malaria in 10 different countries? Data on reasons why individuals not using ITNs (out of all individuals surveyed who own an ITN and are not using it)? Distribution of patients tested for malaria by parasite density Speaker notes How should you present the following data? 1. Line graph 2. Bar Chart 3. Pie Chart 4. Histogram * Summary Make sure that you present your data in a consistent format Use the right graph for the right data and the right audience Label the components of your graphic (title, axis) Indicate source of data and number of observations (n=xx) Add footnote for more explanation Speaker notes In summary, [READ BULLETS] * Creating Graphs Speaker notes Now that we know a little bit about the main types of graphs, we are going to try our hand at making some in Excel. We are including a few helpful hints in this section on more advanced graphing. If you are already very good at making graphs in Excel, please help your neighbors complete the task after you are finished with your work. * Learning objectives Understand basic chart terminology Create charts in PowerPoint using data in Excel Give a description of the data presented in each chart Speaker notes By the end of this session, participants should be able to: [READ BULLETS] * Pie Chart Source: MEASURE Evaluation, Retention, Use and Achievement of “Universal Access” Following the Distribution of Long Lasting Insecticide Treated Nets in Kano State, Nigeria, 2009 Speaker notes Please open the file called graphs from the data presentation folder on your cd. We are going to use the data there to create this and the other charts and graphs in this session. For all of these charts, I want you to try to duplicate the chart shown in the PPT slide exactly. This is not to say that this chart is perfect; however, trying to copy this exactly will allow you to explore some of the chart making functionality in Excel. Go over making this chart with the participants. Show them how to do it using the standard chart layouts in Excel (this is layout 6 in Excel 2007) and also how to adjust aspects such as the legend, data labels and colors of the chart using the layout tab. * Individual Work: Bar Chart Source: Tanzania HIV and Malaria Indicator Survey, 2008 Speaker notes Please now try to create this chart on your own. You may not know how to add the confidence intervals. If that is the case, please finish the other aspects of the chart and I will then give you a demonstration of how to add the CI. They will need to create this chart in excel and export it to PPT. It should look almost exactly like this chart and include the error bars which they will need to be instructed on. Each participant has the data needed to create this chart in an excel file in the folder for this module. * Secondary Axis Speaker notes Please now try to create this chart on your own. If you do not know how to create a secondary Y-axis, please finish the other aspects of the chart and I will then give you a demonstration of how to add the CI. You use secondary axes to be able to chart numbers that have very different scales on the same graph. In this case, there are a lot more malaria cases than deaths. If you charted them on the same axis, you would see a flat line at the bottom for the deaths. They will need to create this chart in excel and export it to PPT. It should look almost exactly like this chart. Each participant has the data needed to create this chart in an excel file in the folder for this module. * Data Interpretation Speaker notes Now that we know how to present our data, we need to be sure that we are interpreting our findings properly. * Analysis vs. Interpretation Analysis: describing data with tables, graphs, or narrative; transforming data into information Interpretation: adding meaning to information by making connections and comparisons and by exploring causes and consequences Speaker notes Analysis is summarizing the data and turning it into information. Data on its own is generally not useful for the decision-making process. Analysis will vary in complexity. Most data analysis is quite simple, but some is much more complicated and requires a great deal of expertise. Interpretation is the process of making sense of the information. What does it mean for your program? * Has the Program Met its Goal? Speaker notes In many cases we need to interpret data to assess the performance of our programs and identify areas that are doing well and others which are underperforming. In this case, our target is to have 80% of children under five sleep under an ITN every night. Have we met our goal? How can you tell? Answer No, the goal has not been met. Country 3 is doing the best but has only reached a little more than half of the goal for ITN. * Interpreting Data Does the indicator meet the target? What is the programmatic relevance of the finding? What are the potential reasons for the finding? How does it compare? (trends, group differences) What other data should be reviewed to understand the finding (triangulation)? Conduct further analysis Speaker notes When interpreting data we may ask these questions: What is the relevance of the unmet target for the program? Is it because we are not meeting our coverage or efficiency goals? Is our quality of care poor? What could be causing this? How are we doing in comparison with other clinics? Districts? What are the potential reasons for the finding? Do data quality issues play a role in what we are observing? What other data should be reviewed to understand the finding (triangulation)? Is there a need to donduct further analysis? * Practical Question: Are ANC clinics in country X reaching their coverage targets for IPTp? Data Source: Routine health information * Speaker notes Now we are going to consider how we could answer the following question: Are ANC clinics reaching their coverage targets for IPTp? We will answer this question using routine health information. Data Source General ANC Registers Which of these variables are relevant to answer your question? Which elements will be included in your numerator and which in your denominator? Answers: 1) New ANC clients, IPTp-1 2) New ANC clients =Denominator, IPTp-1 and IPTp-2= Numerator Code Variables 1. New ANC clients 2. Group pre-test counseled 3. Individual pre-test counseled 4. Accepted HIV test 5A. HIV test result - Positive 5B. HIV test result – Negative 5C. HIV test result - Indeterminate 6 A. Post-test counseled - Positive 6 B. Post-test counseled – Negative 8A. ARV therapy received – Current NVP 9. IPTp-1 10. IPTp-2 * Speaker notes Which of these variables are relevant to answer your question? We’re going to focus on elements 1, 9 and 10. Which elements will be included in your numerator and which in your denominator? IPTp Coverage-Facility Performance Number of ANC clients receiving IPTp Question: Among the five facilities, which one performed better? Answer: Cannot tell because we don’t know the denominators Code Variables Facility 1 Facility 2 Facility 3 Facility 4 Facility 5 9. IPTp-1 536 1435 39 969 862 10. IPTp-2 372 542 38 452 780 Speaker notes Here we have the data on IPTp-1 and 2 to assess facility performance. Among the five facilities, which one performed better? * IPTp Coverage-Facility Performance Number of ANC clients receiving IPTp Question: Now, you have the denominators, which of these facility performed better? Response: Facility 5 Code Variables Facility 1 Facility 2 Facility 3 Facility 4 Facility 5 1 New ANC Clients 744 2708 105 1077 908 9. IPTp-1 536 1435 39 969 862 10. IPTp-2 372 542 38 452 780 Indicator Facility 1 Facility 2 Facility 3 Facility 4 Facility 5 % of new ANC clients who receive IPTp-1 in the past year 72% 53% 37% 90% 95% % of new ANC clients who receive IPTp-2 in the past year 50% 20% 36% 42% 86% Speaker notes Now, you have the denominators, which of these facility performed better? We can see that it was actually facility 5. * Are facilities reaching coverage targets? Target-80% * National coverage target for pregnant women receiving IPTp-2 is 80%. * Speaker notes Here is the same information presented as a chart. We need to use this information to determine, or interpret, whether or not facilites are reaching their coverage targets. Let’s assume that the national coverage target for pregnant women receiving IPTp is 80%. Are the facilities reaching the coverage target? What else can we interpret from this information? Possible answers Facility 1 needs to do a better job following up and increase IPTp coverage a bit. Facility 2 does a better job with IPTp-1 coverage than IPTp-2, but needs to increase coverage of both. Facility 3 does a good job administering IPTp-2 to patients that receive the first round, but they need to increase initial coverage and maintain follow-up. Facility 4 does a good job with IPTp-1 coverage, but this falls of with IPTp-2. Is this loss to follow-up, or are they not administering IPTp-2 when patients return? Facility 5 can be seen as a model and …