Raw Data | Processed Data |

+ readings or measurements taken directly from a measuring instrument (e.g metre rule, stop watch) + readings must be expressed to a fixed number of d.p dictated by the units used and the precision of the instrument. E.g 15.0 cm or 0.150 m when recording the measurements using metre rule. | + readings obtained from computation of the raw data (e.g finding the period of the pendulum) + All processed data must be expressed to a fixed number of d.p or s.f depending on the operation (addition/subtraction & multiplication/division) of the raw data. |

**Commonly used instruments and their precisions**

For example, the instruction in the question paper states to measure 80 ml of water using measuring cylinder. Then, the correct recording is 80.0 ml or 80.0 cm^{3}.

**Table**

For table, a few rules of thumb to follow. Raw data follows the precision of the instrument used. Processed data column follows the s.f of raw data used. It’s fine if your process datas have different s.f. Heading of the table must contain proper unit using solidus. For those more visual learner, you can refer to the picture below.

Some other examples of table (figures below for this section are taken from Handbook for teaching science practical work by Singapore MOE, CPDD, Science unit, Science Branch):

**How many data sets should I collect?**

At least 5 data sets for a best-fit line and at least 8 sets for a best-fit curve.

**Graph plotting**

You can use this acronym to guide your plotting.

**S**: Scale – no awkward scale such as 1:3; 1:7; 1:9 ratio etc.

**L**: (best fit) Line. **Download** this excel file and practice drawing a best-fit line. Rule of thumb: equal number of data points above and below the best fit line. However, it is not always possible.

**A**: Axes labeling (must include unit using solidus if applicable)

**P**: Point (points are plotted accurate to one half of one of the smallest square)

**How do you know if your graph is a best-fit line or best-fit curve? **

It’s useful to read the following parts of the question. If the next part of the question is asking you to find the gradient without specifying exactly at which coordinate, then it has to be best-fit line. Only a line has the same gradient throughout. It might be a smart move to check this out so that you will know how many data points to collect and decide on their range as well.

**Selecting points for calculating gradient**

- Two selected points must be apart for at least 50% of the raw data points. They cannot lie beyond the data points in both directions. A different symbol should be used to indicate these two points, so that marker doesn’t get confused between data points and the chosen points for gradient. It’s possible to choose data point if the data point lies on best-fit line. But you need to remember to read this value to one half of one of the smallest square.
- A triangle tracing should be drawn.
- Their coordinates must be indicated on the line and read to one half of one of the smallest square.

**How should I calculate my gradient?**

Coordinates must be read off to at least one half of one of the smallest square. The value of gradient can be written to 2/3 s.f with unit (if applicable).

**How should I read off my y-intercept from the graph?**

Read off to the accuracy of one half of one of the smallest square.